You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@beam.apache.org by Jan Lukavský <je...@seznam.cz> on 2019/05/21 14:49:38 UTC
Re: Definition of Unified model
Hi,
> Actually, I think it is a larger (open) question whether exactly once
is guaranteed by the model or whether runners are allowed to relax that.
I would think, however, that sources correctly implemented should be
idempotent when run atop an exactly once infrastructure such as Flink of
Dataflow.
I would assume, that the model basically inherits guarantees of
underlying infrastructure. Because Flink does not work as you described
(atomic commit of inputs, state and outputs), but rather a checkpoint
mark is flowing through the DAG much like watermark and on failures
operators are restored and data reprocessed, it (IMHO) implies, that you
have exactly once everywhere in the DAG *but* sinks. That is because
sinks cannot be restored to previous state, instead sinks are supposed
to be idempotent in order for the exactly once to really work (or at
least be able to commit outputs on checkpoint in sink). That implies
that if you don't have sink that is able to commit outputs atomically on
checkpoint, the pipeline execution should be deterministic upon retries,
otherwise shadow writes from failed paths of the pipeline might appear.
Someone from Flink might correct me if I'm wrong, but that's my current
understanding.
> Sounds like we should make this clearer.
I meant that you are right that we must not in any thoughts we are
having forget that streams are by definition out-of-order. That is
property that we cannot change. But - that doesn't limit us from
creating operator that presents the data to UDF as if the stream was
ideally sorted. It can do that by introducing latency, of course.
On 5/21/19 4:01 PM, Robert Bradshaw wrote:
> Reza: One could provide something like this as a utility class, but
> one downside is that it is not scale invariant. It requires a tuning
> parameter that, if to small, won't mitigate the problem, but if to
> big, greatly increases latency. (Possibly one could define a dynamic
> session-like window to solve this though...) It also might be harder
> for runners that *can* cheaply present stuff in timestamp order to
> optimize. (That and, in practice, our annotation-style process methods
> don't lend themselves to easy composition.) I think it could work in
> specific cases though.
>
> More inline below.
>
> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz> wrote:
>> Hi Robert,
>>
>> > Beam has an exactly-once model. If the data was consumed, state
>> mutated, and outputs written downstream (these three are committed
>> together atomically) it will not be replayed. That does not, of course,
>> solve the non-determanism due to ordering (including the fact that two
>> operations reading the same PCollection may view different ordering).
>>
>> I think what you describe is a property of a runner, not of the model,
>> right? I think if I run my pipeline on Flink I will not get this
>> atomicity, because although Flink uses also exactly-once model if might
>> write outputs multiple times.
> Actually, I think it is a larger (open) question whether exactly once
> is guaranteed by the model or whether runners are allowed to relax
> that. I would think, however, that sources correctly implemented
> should be idempotent when run atop an exactly once infrastructure such
> as Flink of Dataflow.
>
>> > 1) Is it correct for a (Stateful)DoFn to assume elements are received
>> in a specific order? In the current model, it is not. Being able to
>> read, handle, and produced out-of-order data, including late data, is a
>> pretty fundamental property of distributed systems.
>>
>> Yes, absolutely. The argument here is not that Stateful ParDo should
>> presume to receive elements in any order, but to _present_ it as such to
>> the user @ProcessElement function.
> Sounds like we should make this clearer.
>
>> > 2) Given that some operations are easier (or possibly only possible)
>> to write when operating on ordered data, and that different runners may
>> have (significantly) cheaper ways to provide this ordering than can be
>> done by the user themselves, should we elevate this to a property of
>> (Stateful?)DoFns that the runner can provide? I think a compelling
>> argument can be made here that we should.
>>
>> +1
>>
>> Jan
>>
>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz> wrote:
>>>> > I don't see batch vs. streaming as part of the model. One can have
>>>> microbatch, or even a runner that alternates between different modes.
>>>>
>>>> Although I understand motivation of this statement, this project name is
>>>> "Apache Beam: An advanced unified programming model". What does the
>>>> model unify, if "streaming vs. batch" is not part of the model?
>>> What I mean is that streaming vs. batch is no longer part of the model
>>> (or ideally API), but pushed down to be a concern of the runner
>>> (executor) of the pipeline.
>>>
>>>
>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>> Hi Kenn,
>>>>
>>>> OK, so if we introduce annotation, we can have stateful ParDo with sorting, that would perfectly resolve my issues. I still have some doubts, though. Let me explain. The current behavior of stateful ParDo has the following properties:
>>>>
>>>> a) might fail in batch, although runs fine in streaming (that is due to the buffering, and unbounded lateness in batch, which was discussed back and forth in this thread)
>>>>
>>>> b) might be non deterministic (this is because the elements arrive at somewhat random order, and even if you do the operation "assign unique ID to elements" this might produce different results when run multiple times)
>>> PCollections are *explicitly* unordered. Any operations that assume or
>>> depend on a specific ordering for correctness (or determinism) must
>>> provide that ordering themselves (i.e. tolerate "arbitrary shuffling
>>> of inputs"). As you point out, that may be very expensive if you have
>>> very hot keys with very large (unbounded) timestamp skew.
>>>
>>> StatefulDoFns are low-level operations that should be used with care;
>>> the simpler windowing model gives determinism in the face of unordered
>>> data (though late data and non-end-of-window triggering introduces
>>> some of the non-determanism back in).
>>>
>>>> What worries me most is the property b), because it seems to me to have serious consequences - not only that if you run twice batch pipeline you would get different results, but even on streaming, when pipeline fails and gets restarted from checkpoint, produced output might differ from the previous run and data from the first run might have already been persisted into sink. That would create somewhat messy outputs.
>>> Beam has an exactly-once model. If the data was consumed, state
>>> mutated, and outputs written downstream (these three are committed
>>> together atomically) it will not be replayed. That does not, of
>>> course, solve the non-determanism due to ordering (including the fact
>>> that two operations reading the same PCollection may view different
>>> ordering).
>>>
>>>> These two properties makes me think that the current implementation is more of a _special case_ than the general one. The general one would be that your state doesn't have the properties to be able to tolerate buffering problems and/or non-determinism. Which is the case where you need sorting in both streaming and batch to be part of the model.
>>>>
>>>> Let me point out one more analogy - that is merging vs. non-merging windows. The general case (merging windows) implies sorting by timestamp in both batch case (explicit) and streaming (buffering). The special case (non-merging windows) doesn't rely on any timestamp ordering, so the sorting and buffering can be dropped. The underlying root cause of this is the same for both stateful ParDo and windowing (essentially, assigning window labels is a stateful operation when windowing function is merging).
>>>>
>>>> The reason for the current behavior of stateful ParDo seems to be performance, but is it right to abandon correctness in favor of performance? Wouldn't it be more consistent to have the default behavior prefer correctness and when you have the specific conditions of state function having special properties, then you can annotate your DoFn (with something like @TimeOrderingAgnostic), which would yield a better performance in that case?
>>> There are two separable questions here.
>>>
>>> 1) Is it correct for a (Stateful)DoFn to assume elements are received
>>> in a specific order? In the current model, it is not. Being able to
>>> read, handle, and produced out-of-order data, including late data, is
>>> a pretty fundamental property of distributed systems.
>>>
>>> 2) Given that some operations are easier (or possibly only possible)
>>> to write when operating on ordered data, and that different runners
>>> may have (significantly) cheaper ways to provide this ordering than
>>> can be done by the user themselves, should we elevate this to a
>>> property of (Stateful?)DoFns that the runner can provide? I think a
>>> compelling argument can be made here that we should.
>>>
>>> - Robert
>>>
>>>
>>>
>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>>>>
>>>> Thanks for the nice small example of a calculation that depends on order. You are right that many state machines have this property. I agree w/ you and Luke that it is convenient for batch processing to sort by event timestamp before running a stateful ParDo. In streaming you could also implement "sort by event timestamp" by buffering until you know all earlier data will be dropped - a slack buffer up to allowed lateness.
>>>>
>>>> I do not think that it is OK to sort in batch and not in streaming. Many state machines diverge very rapidly when things are out of order. So each runner if they see the "@OrderByTimestamp" annotation (or whatever) needs to deliver sorted data (by some mix of buffering and dropping), or to reject the pipeline as unsupported.
>>>>
>>>> And also want to say that this is not the default case - many uses of state & timers in ParDo yield different results at the element level, but the results are equivalent at in the big picture. Such as the example of "assign a unique sequence number to each element" or "group into batches" it doesn't matter exactly what the result is, only that it meets the spec. And other cases like user funnels are monotonic enough that you also don't actually need sorting.
>>>>
>>>> Kenn
>>>>
>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz> wrote:
>>>>> Yes, the problem will arise probably mostly when you have not well distributed keys (or too few keys). I'm really not sure if a pure GBK with a trigger can solve this - it might help to have data driven trigger. There would still be some doubts, though. The main question is still here - people say, that sorting by timestamp before stateful ParDo would be prohibitively slow, but I don't really see why - the sorting is very probably already there. And if not (hash grouping instead of sorted grouping), then the sorting would affect only user defined StatefulParDos.
>>>>>
>>>>> This would suggest that the best way out of this would be really to add annotation, so that the author of the pipeline can decide.
>>>>>
>>>>> If that would be acceptable I think I can try to prepare some basic functionality, but I'm not sure, if I would be able to cover all runners / sdks.
>>>>>
>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>>>>>
>>>>> It is read all per key and window and not just read all (this still won't scale with hot keys in the global window). The GBK preceding the StatefulParDo will guarantee that you are processing all the values for a specific key and window at any given time. Is there a specific window/trigger that is missing that you feel would remove the need for you to use StatefulParDo?
>>>>>
>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz> wrote:
>>>>>> Hi Lukasz,
>>>>>>
>>>>>>> Today, if you must have a strict order, you must guarantee that your StatefulParDo implements the necessary "buffering & sorting" into state.
>>>>>> Yes, no problem with that. But this whole discussion started, because *this doesn't work on batch*. You simply cannot first read everything from distributed storage and then buffer it all into memory, just to read it again, but sorted. That will not work. And even if it would, it would be a terrible waste of resources.
>>>>>>
>>>>>> Jan
>>>>>>
>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>>>>> This discussion brings many really interesting questions for me. :-)
>>>>>>>
>>>>>>> > I don't see batch vs. streaming as part of the model. One can have
>>>>>>> microbatch, or even a runner that alternates between different modes.
>>>>>>>
>>>>>>> Although I understand motivation of this statement, this project name is
>>>>>>> "Apache Beam: An advanced unified programming model". What does the
>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>>>>>>>
>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming are
>>>>>>> exactly the "runtime conditions/characteristics" I refer to. All these
>>>>>>> define several runtime parameters, which in turn define how well/badly
>>>>>>> will the pipeline perform and how many resources might be needed. From
>>>>>>> my point of view, pure streaming should be the most resource demanding
>>>>>>> (if not, why bother with batch? why not run everything in streaming
>>>>>>> only? what will there remain to "unify"?).
>>>>>>>
>>>>>>> > Fortunately, for batch, only the state for a single key needs to be
>>>>>>> preserved at a time, rather than the state for all keys across the range
>>>>>>> of skew. Of course if you have few or hot keys, one can still have
>>>>>>> issues (and this is not specific to StatefulDoFns).
>>>>>>>
>>>>>>> Yes, but here is still the presumption that my stateful DoFn can
>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use case in
>>>>>>> more detail.
>>>>>>>
>>>>>>> Suppose you have input stream consisting of 1s and 0s (and some key for
>>>>>>> each element, which is irrelevant for the demonstration). Your task is
>>>>>>> to calculate in running global window the actual number of changes
>>>>>>> between state 0 and state 1 and vice versa. When the state doesn't
>>>>>>> change, you don't calculate anything. If input (for given key) would be
>>>>>>> (tN denotes timestamp N):
>>>>>>>
>>>>>>> t1: 1
>>>>>>>
>>>>>>> t2: 0
>>>>>>>
>>>>>>> t3: 0
>>>>>>>
>>>>>>> t4: 1
>>>>>>>
>>>>>>> t5: 1
>>>>>>>
>>>>>>> t6: 0
>>>>>>>
>>>>>>> then the output should yield (supposing that default state is zero):
>>>>>>>
>>>>>>> t1: (one: 1, zero: 0)
>>>>>>>
>>>>>>> t2: (one: 1, zero: 1)
>>>>>>>
>>>>>>> t3: (one: 1, zero: 1)
>>>>>>>
>>>>>>> t4: (one: 2, zero: 1)
>>>>>>>
>>>>>>> t5: (one: 2, zero: 1)
>>>>>>>
>>>>>>> t6: (one: 2, zero: 2)
>>>>>>>
>>>>>>> How would you implement this in current Beam semantics?
>>>>>> I think your saying here that I know that my input is ordered in a specific way and since I assume the order when writing my pipeline I can perform this optimization. But there is nothing preventing a runner from noticing that your processing in the global window with a specific type of trigger and re-ordering your inputs/processing to get better performance (since you can't use an AfterWatermark trigger for your pipeline in streaming for the GlobalWindow).
>>>>>>
>>>>>> Today, if you must have a strict order, you must guarantee that your StatefulParDo implements the necessary "buffering & sorting" into state. I can see why you would want an annotation that says I must have timestamp ordered elements, since it makes writing certain StatefulParDos much easier. StatefulParDo is a low-level function, it really is the "here you go and do whatever you need to but here be dragons" function while windowing and triggering is meant to keep many people from writing StatefulParDo in the first place.
>>>>>>
>>>>>>> > Pipelines that fail in the "worst case" batch scenario are likely to
>>>>>>> degrade poorly (possibly catastrophically) when the watermark falls
>>>>>>> behind in streaming mode as well.
>>>>>>>
>>>>>>> But the worst case is defined by input of size (available resources +
>>>>>>> single byte) -> pipeline fail. Although it could have finished, given
>>>>>>> the right conditions.
>>>>>>>
>>>>>>> > This might be reasonable, implemented by default by buffering
>>>>>>> everything and releasing elements as the watermark (+lateness) advances,
>>>>>>> but would likely lead to inefficient (though *maybe* easier to reason
>>>>>>> about) code.
>>>>>>>
>>>>>>> Sure, the pipeline will be less efficient, because it would have to
>>>>>>> buffer and sort the inputs. But at least it will produce correct results
>>>>>>> in cases where updates to state are order-sensitive.
>>>>>>>
>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda (key, values):
>>>>>>> [(key, value) for value in values])?
>>>>>>>
>>>>>>> I'd say roughly yes, but difference would be in the trigger. The trigger
>>>>>>> should ideally fire as soon as watermark (+lateness) crosses element
>>>>>>> with lowest timestamp in the buffer. Although this could be somehow
>>>>>>> emulated by fixed trigger each X millis.
>>>>>>>
>>>>>>> > Or is the underlying desire just to be able to hint to the runner
>>>>>>> that the code may perform better (e.g. require less resources) as skew
>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
>>>>>>>
>>>>>>> No, the sorting would have to be done in streaming case as well. That is
>>>>>>> an imperative of the unified model. I think it is possible to sort by
>>>>>>> timestamp only in batch case (and do it for *all* batch stateful pardos
>>>>>>> without annotation), or introduce annotation, but then make the same
>>>>>>> guarantees for streaming case as well.
>>>>>>>
>>>>>>> Jan
>>>>>>>
>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský <je...@seznam.cz> wrote:
>>>>>>>>> Hi Robert,
>>>>>>>>>
>>>>>>>>> yes, I think you rephrased my point - although no *explicit* guarantees
>>>>>>>>> of ordering are given in either mode, there is *implicit* ordering in
>>>>>>>>> streaming case that is due to nature of the processing - the difference
>>>>>>>>> between watermark and timestamp of elements flowing through the pipeline
>>>>>>>>> are generally low (too high difference leads to the overbuffering
>>>>>>>>> problem), but there is no such bound on batch.
>>>>>>>> Fortunately, for batch, only the state for a single key needs to be
>>>>>>>> preserved at a time, rather than the state for all keys across the
>>>>>>>> range of skew. Of course if you have few or hot keys, one can still
>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>>>>>>>>
>>>>>>>>> As a result, I see a few possible solutions:
>>>>>>>>>
>>>>>>>>> - the best and most natural seems to be extension of the model, so
>>>>>>>>> that it defines batch as not only "streaming pipeline executed in batch
>>>>>>>>> fashion", but "pipeline with at least as good runtime characteristics as
>>>>>>>>> in streaming case, executed in batch fashion", I really don't think that
>>>>>>>>> there are any conflicts with the current model, or that this could
>>>>>>>>> affect performance, because the required sorting (as pointed by
>>>>>>>>> Aljoscha) is very probably already done during translation of stateful
>>>>>>>>> pardos. Also note that this definition only affects user defined
>>>>>>>>> stateful pardos
>>>>>>>> I don't see batch vs. streaming as part of the model. One can have
>>>>>>>> microbatch, or even a runner that alternates between different modes.
>>>>>>>> The model describes what the valid outputs are given a (sometimes
>>>>>>>> partial) set of inputs. It becomes really hard to define things like
>>>>>>>> "as good runtime characteristics." Once you allow any
>>>>>>>> out-of-orderedness, it is not very feasible to try and define (and
>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>>>>>>>> out-of-orderedness.
>>>>>>>>
>>>>>>>> Pipelines that fail in the "worst case" batch scenario are likely to
>>>>>>>> degrade poorly (possibly catastrophically) when the watermark falls
>>>>>>>> behind in streaming mode as well.
>>>>>>>>
>>>>>>>>> - another option would be to introduce annotation for DoFns (e.g.
>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in the sorting
>>>>>>>>> in batch case - but - this extension would have to ensure the sorting in
>>>>>>>>> streaming mode also - it would require definition of allowed lateness,
>>>>>>>>> and triggger (essentially similar to window)
>>>>>>>> This might be reasonable, implemented by default by buffering
>>>>>>>> everything and releasing elements as the watermark (+lateness)
>>>>>>>> advances, but would likely lead to inefficient (though *maybe* easier
>>>>>>>> to reason about) code. Not sure about the semantics of triggering
>>>>>>>> here, especially data-driven triggers. Would it be roughly equivalent
>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for value in
>>>>>>>> values])?
>>>>>>>>
>>>>>>>> Or is the underlying desire just to be able to hint to the runner that
>>>>>>>> the code may perform better (e.g. require less resources) as skew is
>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>>>>>>>>
>>>>>>>>> - last option would be to introduce these "higher order guarantees" in
>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the worst
>>>>>>>>> option to me
>>>>>>>>>
>>>>>>>>> I see the first two options quite equally good, although the letter one
>>>>>>>>> is probably more time consuming to implement. But it would bring
>>>>>>>>> additional feature to streaming case as well.
>>>>>>>>>
>>>>>>>>> Thanks for any thoughts.
>>>>>>>>>
>>>>>>>>> Jan
>>>>>>>>>
>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský <je...@seznam.cz> wrote:
>>>>>>>>>>> Hi Reuven,
>>>>>>>>>>>
>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch runners.
>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside the state works for absolutely unbounded out-of-orderness of elements. That basically (practically) can work only for cases, where the order of input elements doesn't matter. But, "state" can refer to "state machine", and any time you have a state machine involved, then the ordering of elements would matter.
>>>>>>>>>> No guarantees on order are provided in *either* streaming or batch
>>>>>>>>>> mode by the model. However, it is the case that in order to make
>>>>>>>>>> forward progress most streaming runners attempt to limit the amount of
>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs. processing
>>>>>>>>>> time) to make forward progress, which in turn could help cap the
>>>>>>>>>> amount of state that must be held concurrently, whereas a batch runner
>>>>>>>>>> may not allow any state to be safely discarded until the whole
>>>>>>>>>> timeline from infinite past to infinite future has been observed.
>>>>>>>>>>
>>>>>>>>>> Also, as pointed out, state is not preserved "batch to batch" in batch mode.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels <mx...@apache.org> wrote:
>>>>>>>>>>
>>>>>>>>>>>> batch semantics and streaming semantics differs only in that I can have GlobalWindow with default trigger on batch and cannot on stream
>>>>>>>>>>> You can have a GlobalWindow in streaming with a default trigger. You
>>>>>>>>>>> could define additional triggers that do early firings. And you could
>>>>>>>>>>> even trigger the global window by advancing the watermark to +inf.
>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with default
>>>>>>>>>> trigger on unbounded PCollections in the SDK because this is more
>>>>>>>>>> likely to be user error than an actual desire to have no output until
>>>>>>>>>> drain. But it's semantically valid in the model.
Re: Definition of Unified model
Posted by Jan Lukavský <je...@seznam.cz>.
That's right, but is there a filesystem, that allows unbounded size of
files? If there will always be an upper size limit, does that mean that
you cannot use the order of elements in the file as is? You might need
to transfer the offset from one file to another (that's how Kafka does
it), but that implies that you don't use what natively gives you the
batch storage, but you store the offset yourself (as metadata).
Either way, maybe the discussion is not that important, because the
invariant requirement persists - there has to be a sequential observer
of the data, that creates sequence of updates in the order the data was
observed and persists this order. If you have two observers of data,
each storing his own (even unbounded in size) file, then (if partition
by key is not enforced) I'd say the ordering cannot be used.
This mechanism seems to me related to what limits parallelism in
streaming sources and why batch sources are generally better parallelised.
Jan
On 5/30/19 1:35 PM, Reuven Lax wrote:
> Files can grow (depending on the filesystem), and tailing growing
> files is a valid use case.
>
> On Wed, May 29, 2019 at 3:23 PM Jan Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
>
> > Offsets within a file, unordered between files seems exactly
> analogous with offsets within a partition, unordered between
> partitions,
> right?
>
> Not exactly. The key difference is in that partitions in streaming
> stores are defined (on purpose, and with key impact on this
> discussion)
> as unbounded sequence of appends. Files, on the other hand are
> always of
> finite size. This difference makes the semantics of offsets in
> partitioned stream useful, because the are guaranteed to only
> increase.
> On batch stores as files, these offsets would have to start from zero
> after some (finite) time, which makes them useless for comparison.
>
> On 5/29/19 2:44 PM, Robert Bradshaw wrote:
> > On Tue, May 28, 2019 at 12:18 PM Jan Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
> >> As I understood it, Kenn was supporting the idea that sequence
> metadata
> >> is preferable over FIFO. I was trying to point out, that it
> even should
> >> provide the same functionally as FIFO, plus one important more -
> >> reproducibility and ability to being persisted and reused the
> same way
> >> in batch and streaming.
> >>
> >> There is no doubt, that sequence metadata can be stored in every
> >> storage. But, regarding some implicit ordering that sources
> might have -
> >> yes, of course, data written into HDFS or Cloud Storage has
> ordering,
> >> but only partial - inside some bulk (e.g. file) and the
> ordering is not
> >> defined correctly on boundaries of these bulks (between files).
> That is
> >> why I'd say, that ordering of sources is relevant only for
> >> (partitioned!) streaming sources and generally always reduces to
> >> sequence metadata (e.g. offsets).
> > Offsets within a file, unordered between files seems exactly
> analogous
> > with offsets within a partition, unordered between partitions,
> right?
> >
> >> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
> >>> Huge +1 to all Kenn said.
> >>>
> >>> Jan, batch sources can have orderings too, just like Kafka. I
> think
> >>> it's reasonable (for both batch and streaming) that if a
> source has an
> >>> ordering that is an important part of the data, it should preserve
> >>> this ordering into the data itself (e.g. as sequence numbers,
> offsets,
> >>> etc.)
> >>>
> >>> On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles
> <kenn@apache.org <ma...@apache.org>> wrote:
> >>>> I strongly prefer explicit sequence metadata over FIFO
> requirements, because:
> >>>>
> >>>> - FIFO is complex to specify: for example Dataflow has
> "per stage key-to-key" FIFO today, but it is not guaranteed to
> remain so (plus "stage" is not a portable concept, nor even
> guaranteed to remain a Dataflow concept)
> >>>> - complex specifications are by definition poor usability
> (if necessary, then it is what it is)
> >>>> - overly restricts the runner, reduces parallelism, for
> example any non-stateful ParDo has per-element parallelism, not
> per "key"
> >>>> - another perspective on that: FIFO makes everyone pay
> rather than just the transform that requires exactly sequencing
> >>>> - previous implementation details like reshuffles become
> part of the model
> >>>> - I'm not even convinced the use cases involved are
> addressed by some careful FIFO restrictions; many sinks re-key and
> they would all have to become aware of how keying of a sequence of
> "stages" affects the end-to-end FIFO
> >>>>
> >>>> A noop becoming a non-noop is essentially the mathematical
> definition of moving from higher-level to lower-level abstraction.
> >>>>
> >>>> So this strikes at the core question of what level of
> abstraction Beam aims to represent. Lower-level means there are
> fewer possible implementations and it is more tied to the
> underlying architecture, and anything not near-exact match pays a
> huge penalty. Higher-level means there are more implementations
> possible with different tradeoffs, though they may all pay a minor
> penalty.
> >>>>
> >>>> I could be convinced to change my mind, but it needs some
> extensive design, examples, etc. I think it is probably about the
> most consequential design decision in the whole Beam model, around
> the same level as the decision to use ParDo and GBK as the
> primitives IMO.
> >>>>
> >>>> Kenn
>
Re: Definition of Unified model
Posted by Reuven Lax <re...@google.com>.
Files can grow (depending on the filesystem), and tailing growing files is
a valid use case.
On Wed, May 29, 2019 at 3:23 PM Jan Lukavský <je...@seznam.cz> wrote:
> > Offsets within a file, unordered between files seems exactly
> analogous with offsets within a partition, unordered between partitions,
> right?
>
> Not exactly. The key difference is in that partitions in streaming
> stores are defined (on purpose, and with key impact on this discussion)
> as unbounded sequence of appends. Files, on the other hand are always of
> finite size. This difference makes the semantics of offsets in
> partitioned stream useful, because the are guaranteed to only increase.
> On batch stores as files, these offsets would have to start from zero
> after some (finite) time, which makes them useless for comparison.
>
> On 5/29/19 2:44 PM, Robert Bradshaw wrote:
> > On Tue, May 28, 2019 at 12:18 PM Jan Lukavský <je...@seznam.cz> wrote:
> >> As I understood it, Kenn was supporting the idea that sequence metadata
> >> is preferable over FIFO. I was trying to point out, that it even should
> >> provide the same functionally as FIFO, plus one important more -
> >> reproducibility and ability to being persisted and reused the same way
> >> in batch and streaming.
> >>
> >> There is no doubt, that sequence metadata can be stored in every
> >> storage. But, regarding some implicit ordering that sources might have -
> >> yes, of course, data written into HDFS or Cloud Storage has ordering,
> >> but only partial - inside some bulk (e.g. file) and the ordering is not
> >> defined correctly on boundaries of these bulks (between files). That is
> >> why I'd say, that ordering of sources is relevant only for
> >> (partitioned!) streaming sources and generally always reduces to
> >> sequence metadata (e.g. offsets).
> > Offsets within a file, unordered between files seems exactly analogous
> > with offsets within a partition, unordered between partitions, right?
> >
> >> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
> >>> Huge +1 to all Kenn said.
> >>>
> >>> Jan, batch sources can have orderings too, just like Kafka. I think
> >>> it's reasonable (for both batch and streaming) that if a source has an
> >>> ordering that is an important part of the data, it should preserve
> >>> this ordering into the data itself (e.g. as sequence numbers, offsets,
> >>> etc.)
> >>>
> >>> On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org>
> wrote:
> >>>> I strongly prefer explicit sequence metadata over FIFO requirements,
> because:
> >>>>
> >>>> - FIFO is complex to specify: for example Dataflow has "per stage
> key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage"
> is not a portable concept, nor even guaranteed to remain a Dataflow concept)
> >>>> - complex specifications are by definition poor usability (if
> necessary, then it is what it is)
> >>>> - overly restricts the runner, reduces parallelism, for example
> any non-stateful ParDo has per-element parallelism, not per "key"
> >>>> - another perspective on that: FIFO makes everyone pay rather than
> just the transform that requires exactly sequencing
> >>>> - previous implementation details like reshuffles become part of
> the model
> >>>> - I'm not even convinced the use cases involved are addressed by
> some careful FIFO restrictions; many sinks re-key and they would all have
> to become aware of how keying of a sequence of "stages" affects the
> end-to-end FIFO
> >>>>
> >>>> A noop becoming a non-noop is essentially the mathematical definition
> of moving from higher-level to lower-level abstraction.
> >>>>
> >>>> So this strikes at the core question of what level of abstraction
> Beam aims to represent. Lower-level means there are fewer possible
> implementations and it is more tied to the underlying architecture, and
> anything not near-exact match pays a huge penalty. Higher-level means there
> are more implementations possible with different tradeoffs, though they may
> all pay a minor penalty.
> >>>>
> >>>> I could be convinced to change my mind, but it needs some extensive
> design, examples, etc. I think it is probably about the most consequential
> design decision in the whole Beam model, around the same level as the
> decision to use ParDo and GBK as the primitives IMO.
> >>>>
> >>>> Kenn
>
Re: Definition of Unified model
Posted by Jan Lukavský <je...@seznam.cz>.
> Offsets within a file, unordered between files seems exactly
analogous with offsets within a partition, unordered between partitions,
right?
Not exactly. The key difference is in that partitions in streaming
stores are defined (on purpose, and with key impact on this discussion)
as unbounded sequence of appends. Files, on the other hand are always of
finite size. This difference makes the semantics of offsets in
partitioned stream useful, because the are guaranteed to only increase.
On batch stores as files, these offsets would have to start from zero
after some (finite) time, which makes them useless for comparison.
On 5/29/19 2:44 PM, Robert Bradshaw wrote:
> On Tue, May 28, 2019 at 12:18 PM Jan Lukavský <je...@seznam.cz> wrote:
>> As I understood it, Kenn was supporting the idea that sequence metadata
>> is preferable over FIFO. I was trying to point out, that it even should
>> provide the same functionally as FIFO, plus one important more -
>> reproducibility and ability to being persisted and reused the same way
>> in batch and streaming.
>>
>> There is no doubt, that sequence metadata can be stored in every
>> storage. But, regarding some implicit ordering that sources might have -
>> yes, of course, data written into HDFS or Cloud Storage has ordering,
>> but only partial - inside some bulk (e.g. file) and the ordering is not
>> defined correctly on boundaries of these bulks (between files). That is
>> why I'd say, that ordering of sources is relevant only for
>> (partitioned!) streaming sources and generally always reduces to
>> sequence metadata (e.g. offsets).
> Offsets within a file, unordered between files seems exactly analogous
> with offsets within a partition, unordered between partitions, right?
>
>> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
>>> Huge +1 to all Kenn said.
>>>
>>> Jan, batch sources can have orderings too, just like Kafka. I think
>>> it's reasonable (for both batch and streaming) that if a source has an
>>> ordering that is an important part of the data, it should preserve
>>> this ordering into the data itself (e.g. as sequence numbers, offsets,
>>> etc.)
>>>
>>> On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org> wrote:
>>>> I strongly prefer explicit sequence metadata over FIFO requirements, because:
>>>>
>>>> - FIFO is complex to specify: for example Dataflow has "per stage key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage" is not a portable concept, nor even guaranteed to remain a Dataflow concept)
>>>> - complex specifications are by definition poor usability (if necessary, then it is what it is)
>>>> - overly restricts the runner, reduces parallelism, for example any non-stateful ParDo has per-element parallelism, not per "key"
>>>> - another perspective on that: FIFO makes everyone pay rather than just the transform that requires exactly sequencing
>>>> - previous implementation details like reshuffles become part of the model
>>>> - I'm not even convinced the use cases involved are addressed by some careful FIFO restrictions; many sinks re-key and they would all have to become aware of how keying of a sequence of "stages" affects the end-to-end FIFO
>>>>
>>>> A noop becoming a non-noop is essentially the mathematical definition of moving from higher-level to lower-level abstraction.
>>>>
>>>> So this strikes at the core question of what level of abstraction Beam aims to represent. Lower-level means there are fewer possible implementations and it is more tied to the underlying architecture, and anything not near-exact match pays a huge penalty. Higher-level means there are more implementations possible with different tradeoffs, though they may all pay a minor penalty.
>>>>
>>>> I could be convinced to change my mind, but it needs some extensive design, examples, etc. I think it is probably about the most consequential design decision in the whole Beam model, around the same level as the decision to use ParDo and GBK as the primitives IMO.
>>>>
>>>> Kenn
Re: Definition of Unified model
Posted by Dan Sotolongo <ds...@google.com>.
Jan, I believe what Luke is suggesting is essentially the model implemented
by Timely Dataflow <https://github.com/TimelyDataflow/timely-dataflow>.
It's pretty interesting reading.
On Thu, May 30, 2019 at 8:04 AM Lukasz Cwik <lc...@google.com> wrote:
> You don't need the ordering, you need support for higher dimensionality in
> the element tracking so that a runner can track that it is making forward
> progress, schedule and execute timers, know when side inputs are ready, ...
>
> On Thu, May 30, 2019 at 4:06 AM Jan Lukavský <je...@seznam.cz> wrote:
>
>> Hi Lukasz, do you suggest, that ordering can have some relation to
>> iterations? That sounds interesting, and from the top of my head, it can be
>> true - you have to order on iteration number to get consistent results.
>> Maybe we are tracking something deeper here! :-)
>>
>> Jan
>> On 5/29/19 7:02 PM, Lukasz Cwik wrote:
>>
>> Expanding the dimensionality could be the basis for loops within the
>> graph since loops could be modeled as (time, loop iteration #, nested loop
>> iteration #, nested nested loop iteration #, ...)
>>
>> On Tue, May 28, 2019 at 12:10 PM Jan Lukavský <je...@seznam.cz> wrote:
>>
>>> Could this be solved by "expanding the dimensionality" of time field?
>>> What I mean by that - if input element to to FlatMap has sequence number T,
>>> then the (stateless) FlatMap knows the ordering of output elements, right?
>>> If it would expand the field by which it will next sort the elements to (X,
>>> 1), (X, 2), ... (X, N), then it would be possible to sort the elements back
>>> later. There seems to be no need for state to achieve that, or?
>>>
>>> Jan
>>> On 5/28/19 6:52 PM, Reuven Lax wrote:
>>>
>>> A slightly larger concern: it also will force users to create stateful
>>> DoFns everywhere to generate these sequence numbers. If I have a ParDo that
>>> is not a simple 1:1 transform (i.e. not MapElements), then the ParDo will
>>> need to generate its own sequence numbers for ordering, and the only safe
>>> way to do so is to use a stateful DoFn. This turns what used to be a simple
>>> in-memory DoFn into one that has to access state. Also I believe many
>>> runners will not fuse stateful DoFns. While none of this poses a problem
>>> for the model, it could make ordering extremely expensive to achieve.
>>>
>>> Reuven
>>>
>>> On Tue, May 28, 2019 at 6:09 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>
>>>> Hi Reuven,
>>>>
>>>> > It also gets awkward with Flatten - the sequence number is no longer
>>>> enough, you must also encode which side of the flatten each element came
>>>> from.
>>>>
>>>> That is a generic need. Even if you read data from Kafka, the offsets
>>>> are comparable only inside single partition. So, for Kafka to work as a
>>>> FIFO for ordering, elements with same key have to be pushed to the same
>>>> partition (otherwise Kafka cannot act as FIFO, because different partitions
>>>> can be handled by different brokers, which means different observers and
>>>> they therefore might not agree on the order of events). So if we want to
>>>> emulate FIFO per key, then the sequence IDs have also be per key.
>>>> On 5/28/19 2:33 PM, Reuven Lax wrote:
>>>>
>>>> Sequence metadata does have the disadvantage that users can no longer
>>>> use the types coming from the source. You must create a new type that
>>>> contains a sequence number (unless Beam provides this). It also gets
>>>> awkward with Flatten - the sequence number is no longer enough, you must
>>>> also encode which side of the flatten each element came from.
>>>>
>>>> On Tue, May 28, 2019 at 3:18 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>>
>>>>> As I understood it, Kenn was supporting the idea that sequence
>>>>> metadata
>>>>> is preferable over FIFO. I was trying to point out, that it even
>>>>> should
>>>>> provide the same functionally as FIFO, plus one important more -
>>>>> reproducibility and ability to being persisted and reused the same way
>>>>> in batch and streaming.
>>>>>
>>>>> There is no doubt, that sequence metadata can be stored in every
>>>>> storage. But, regarding some implicit ordering that sources might have
>>>>> -
>>>>> yes, of course, data written into HDFS or Cloud Storage has ordering,
>>>>> but only partial - inside some bulk (e.g. file) and the ordering is
>>>>> not
>>>>> defined correctly on boundaries of these bulks (between files). That
>>>>> is
>>>>> why I'd say, that ordering of sources is relevant only for
>>>>> (partitioned!) streaming sources and generally always reduces to
>>>>> sequence metadata (e.g. offsets).
>>>>>
>>>>> Jan
>>>>>
>>>>> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
>>>>> > Huge +1 to all Kenn said.
>>>>> >
>>>>> > Jan, batch sources can have orderings too, just like Kafka. I think
>>>>> > it's reasonable (for both batch and streaming) that if a source has
>>>>> an
>>>>> > ordering that is an important part of the data, it should preserve
>>>>> > this ordering into the data itself (e.g. as sequence numbers,
>>>>> offsets,
>>>>> > etc.)
>>>>> >
>>>>> > On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org>
>>>>> wrote:
>>>>> >> I strongly prefer explicit sequence metadata over FIFO
>>>>> requirements, because:
>>>>> >>
>>>>> >> - FIFO is complex to specify: for example Dataflow has "per stage
>>>>> key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage"
>>>>> is not a portable concept, nor even guaranteed to remain a Dataflow concept)
>>>>> >> - complex specifications are by definition poor usability (if
>>>>> necessary, then it is what it is)
>>>>> >> - overly restricts the runner, reduces parallelism, for example
>>>>> any non-stateful ParDo has per-element parallelism, not per "key"
>>>>> >> - another perspective on that: FIFO makes everyone pay rather
>>>>> than just the transform that requires exactly sequencing
>>>>> >> - previous implementation details like reshuffles become part of
>>>>> the model
>>>>> >> - I'm not even convinced the use cases involved are addressed by
>>>>> some careful FIFO restrictions; many sinks re-key and they would all have
>>>>> to become aware of how keying of a sequence of "stages" affects the
>>>>> end-to-end FIFO
>>>>> >>
>>>>> >> A noop becoming a non-noop is essentially the mathematical
>>>>> definition of moving from higher-level to lower-level abstraction.
>>>>> >>
>>>>> >> So this strikes at the core question of what level of abstraction
>>>>> Beam aims to represent. Lower-level means there are fewer possible
>>>>> implementations and it is more tied to the underlying architecture, and
>>>>> anything not near-exact match pays a huge penalty. Higher-level means there
>>>>> are more implementations possible with different tradeoffs, though they may
>>>>> all pay a minor penalty.
>>>>> >>
>>>>> >> I could be convinced to change my mind, but it needs some extensive
>>>>> design, examples, etc. I think it is probably about the most consequential
>>>>> design decision in the whole Beam model, around the same level as the
>>>>> decision to use ParDo and GBK as the primitives IMO.
>>>>> >>
>>>>> >> Kenn
>>>>> >>
>>>>> >> On Thu, May 23, 2019 at 10:17 AM Reuven Lax <re...@google.com>
>>>>> wrote:
>>>>> >>> Not really. I'm suggesting that some variant of FIFO ordering is
>>>>> necessary, which requires either runners natively support FIFO ordering or
>>>>> transforms adding some extra sequence number to each record to sort by.
>>>>> >>>
>>>>> >>> I still think your proposal is very useful by the way. I'm merely
>>>>> pointing out that to solve the state-machine problem we probably need
>>>>> something more.
>>>>> >>>
>>>>> >>> Reuven
>>>>> >>>
>>>>> >>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je...@seznam.cz>
>>>>> wrote:
>>>>> >>>> Hi,
>>>>> >>>> yes. It seems that ordering by user supplied UDF makes sense and
>>>>> I will update the design proposal accordingly.
>>>>> >>>> Would that solve the issues you mention?
>>>>> >>>> Jan
>>>>> >>>> ---------- Původní e-mail ----------
>>>>> >>>> Od: Reuven Lax <re...@google.com>
>>>>> >>>> Komu: dev <de...@beam.apache.org>
>>>>> >>>> Datum: 23. 5. 2019 18:44:38
>>>>> >>>> Předmět: Re: Definition of Unified model
>>>>> >>>>
>>>>> >>>> I'm simply saying that timestamp ordering is insufficient for
>>>>> state machines. I wasn't proposing Kafka as a solution - that was simply an
>>>>> example of how people solve this problem in other scenarios.
>>>>> >>>>
>>>>> >>>> BTW another example of ordering: Imagine today that you have a
>>>>> triggered Sum aggregation writing out to a key-value sink. In theory we
>>>>> provide no ordering, so the sink might write the triggered sums in the
>>>>> wrong order, ending up with an incorrect value in the sink. In this case
>>>>> you probably want values ordered by trigger pane index.
>>>>> >>>>
>>>>> >>>> Reuven
>>>>> >>>>
>>>>> >>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz>
>>>>> wrote:
>>>>> >>>>
>>>>> >>>> Hi Reuven,
>>>>> >>>> I share the view point of Robert. I think the isuue you refer to
>>>>> is not in reality related to timestamps, but to the fact, that ordering of
>>>>> events in time is observer dependent (either caused by relativity, or time
>>>>> skew, essentially this has the same consequences). And the resolution in
>>>>> fact isn't Kafka, but generally an authoritative observer, that tells you
>>>>> "I saw the events in this order". And you either have one (and have the
>>>>> outcome of his observation persisted in the data - e.g. as offset in Kafka
>>>>> partition), then you should be able to use it (maybe that suggests afterall
>>>>> that sorting by some user supplied UDF might make sense), or do not have
>>>>> it, and then any interpretation of the data seems to be equally valid.
>>>>> Although determinism is fine, of course.
>>>>> >>>> Jan
>>>>> >>>> ---------- Původní e-mail ----------
>>>>> >>>> Od: Reuven Lax <re...@google.com>
>>>>> >>>> Komu: dev <de...@beam.apache.org>
>>>>> >>>> Datum: 23. 5. 2019 17:39:12
>>>>> >>>> Předmět: Re: Definition of Unified model
>>>>> >>>>
>>>>> >>>> So an example would be elements of type "startUserSession" and
>>>>> "endUserSession" (website sessions, not Beam sessions). Logically you may
>>>>> need to process them in the correct order if you have any sort of
>>>>> state-machine logic. However timestamp ordering is never guaranteed to
>>>>> match the logical ordering. Not only might you have several elements with
>>>>> the same timestamp, but in reality time skew across backend servers can
>>>>> cause the events to have timestamps in reverse order of the actual
>>>>> causality order.
>>>>> >>>>
>>>>> >>>> People do solve this problem today though. Publish the events to
>>>>> Kafka, making sure that events for the same user end up in the same Kafka
>>>>> partition. This ensures that the events appear in the Kafka partitions in
>>>>> causality order, even if the timestamp order doesn't match. The your Kafka
>>>>> subscriber simply process the elements in each partition in order.
>>>>> >>>>
>>>>> >>>> I think the ability to impose FIFO causality ordering is what's
>>>>> needed for any state-machine work. Timestamp ordering has advantages
>>>>> (though often I think the advantage is in state), but does not solve this
>>>>> problem.
>>>>> >>>>
>>>>> >>>> Reuven
>>>>> >>>>
>>>>> >>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <
>>>>> robertwb@google.com> wrote:
>>>>> >>>>
>>>>> >>>> Good point.
>>>>> >>>>
>>>>> >>>> The "implementation-specific" way I would do this is
>>>>> >>>> window-by-instant, followed by a DoFn that gets all the elements
>>>>> with
>>>>> >>>> the same timestamp and sorts/acts accordingly, but this counts on
>>>>> the
>>>>> >>>> runner producing windows in timestamp order (likely?) and also the
>>>>> >>>> subsequent DoFn getting them in this order (also likely, due to
>>>>> >>>> fusion).
>>>>> >>>>
>>>>> >>>> One could make the argument that, though it does not provide
>>>>> >>>> deterministic behavior, getting elements of the same timestamp in
>>>>> >>>> different orders should produce equally valid interpretations of
>>>>> the
>>>>> >>>> data. (After all, due to relatively, timestamps are not
>>>>> technically
>>>>> >>>> well ordered across space.) I can see how data-dependent
>>>>> tiebreakers
>>>>> >>>> could be useful, or promises of preservation of order between
>>>>> >>>> operations.
>>>>> >>>>
>>>>> >>>> - Robert
>>>>> >>>>
>>>>> >>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com>
>>>>> wrote:
>>>>> >>>>> So Jan's example of state machines is quite a valid use case for
>>>>> ordering. However in my experience, timestamp ordering is insufficient for
>>>>> state machines. Elements that cause state transitions might come in with
>>>>> the exact same timestamp, yet still have a necessary ordering. Especially
>>>>> given Beam's decision to have milliseconds timestamps this is possible, but
>>>>> even at microsecond or nanosecond precision this can happen at scale. To
>>>>> handle state machines you usually need some sort of FIFO ordering along
>>>>> with an ordered sources, such as Kafka, not timestamp ordering.
>>>>> >>>>>
>>>>> >>>>> Reuven
>>>>> >>>>>
>>>>> >>>>> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz>
>>>>> wrote:
>>>>> >>>>>> Hi all,
>>>>> >>>>>>
>>>>> >>>>>> thanks everyone for this discussion. I think I have gathered
>>>>> enough
>>>>> >>>>>> feedback to be able to put down a proposition for changes,
>>>>> which I will
>>>>> >>>>>> do and send to this list for further discussion. There are
>>>>> still doubts
>>>>> >>>>>> remaining the non-determinism and it's relation to outputs
>>>>> stability vs.
>>>>> >>>>>> latency. But I will try to clarify all this in the design
>>>>> document.
>>>>> >>>>>>
>>>>> >>>>>> Thanks,
>>>>> >>>>>>
>>>>> >>>>>> Jan
>>>>> >>>>>>
>>>>> >>>>>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>>>>> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's
>>>>> my
>>>>> >>>>>>>> current understanding.
>>>>> >>>>>>> In essence your description of how exactly-once works in Flink
>>>>> is
>>>>> >>>>>>> correct. The general assumption in Flink is that pipelines
>>>>> must be
>>>>> >>>>>>> deterministic and thus produce idempotent writes in the case of
>>>>> >>>>>>> failures. However, that doesn't mean Beam sinks can't
>>>>> guarantee a bit
>>>>> >>>>>>> more with what Flink has to offer.
>>>>> >>>>>>>
>>>>> >>>>>>> Luke already mentioned the design discussions for
>>>>> @RequiresStableInput
>>>>> >>>>>>> which ensures idempotent writes for non-deterministic
>>>>> pipelines. This
>>>>> >>>>>>> is not part of the model but an optional Beam feature.
>>>>> >>>>>>>
>>>>> >>>>>>> We recently implemented support for @RequiresStableInput in
>>>>> the Flink
>>>>> >>>>>>> Runner. Reuven mentioned the Flink checkpoint confirmation,
>>>>> which
>>>>> >>>>>>> allows us to buffer (and checkpoint) processed data and only
>>>>> emit it
>>>>> >>>>>>> once a Flink checkpoint has completed.
>>>>> >>>>>>>
>>>>> >>>>>>> Cheers,
>>>>> >>>>>>> Max
>>>>> >>>>>>>
>>>>> >>>>>>> On 21.05.19 16:49, Jan Lukavský wrote:
>>>>> >>>>>>>> Hi,
>>>>> >>>>>>>>
>>>>> >>>>>>>> > Actually, I think it is a larger (open) question whether
>>>>> exactly
>>>>> >>>>>>>> once is guaranteed by the model or whether runners are
>>>>> allowed to
>>>>> >>>>>>>> relax that. I would think, however, that sources correctly
>>>>> >>>>>>>> implemented should be idempotent when run atop an exactly once
>>>>> >>>>>>>> infrastructure such as Flink of Dataflow.
>>>>> >>>>>>>>
>>>>> >>>>>>>> I would assume, that the model basically inherits guarantees
>>>>> of
>>>>> >>>>>>>> underlying infrastructure. Because Flink does not work as you
>>>>> >>>>>>>> described (atomic commit of inputs, state and outputs), but
>>>>> rather a
>>>>> >>>>>>>> checkpoint mark is flowing through the DAG much like
>>>>> watermark and on
>>>>> >>>>>>>> failures operators are restored and data reprocessed, it
>>>>> (IMHO)
>>>>> >>>>>>>> implies, that you have exactly once everywhere in the DAG
>>>>> *but*
>>>>> >>>>>>>> sinks. That is because sinks cannot be restored to previous
>>>>> state,
>>>>> >>>>>>>> instead sinks are supposed to be idempotent in order for the
>>>>> exactly
>>>>> >>>>>>>> once to really work (or at least be able to commit outputs on
>>>>> >>>>>>>> checkpoint in sink). That implies that if you don't have sink
>>>>> that is
>>>>> >>>>>>>> able to commit outputs atomically on checkpoint, the pipeline
>>>>> >>>>>>>> execution should be deterministic upon retries, otherwise
>>>>> shadow
>>>>> >>>>>>>> writes from failed paths of the pipeline might appear.
>>>>> >>>>>>>>
>>>>> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's
>>>>> my
>>>>> >>>>>>>> current understanding.
>>>>> >>>>>>>>
>>>>> >>>>>>>> > Sounds like we should make this clearer.
>>>>> >>>>>>>>
>>>>> >>>>>>>> I meant that you are right that we must not in any thoughts
>>>>> we are
>>>>> >>>>>>>> having forget that streams are by definition out-of-order.
>>>>> That is
>>>>> >>>>>>>> property that we cannot change. But - that doesn't limit us
>>>>> from
>>>>> >>>>>>>> creating operator that presents the data to UDF as if the
>>>>> stream was
>>>>> >>>>>>>> ideally sorted. It can do that by introducing latency, of
>>>>> course.
>>>>> >>>>>>>>
>>>>> >>>>>>>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>>>>> >>>>>>>>> Reza: One could provide something like this as a utility
>>>>> class, but
>>>>> >>>>>>>>> one downside is that it is not scale invariant. It requires
>>>>> a tuning
>>>>> >>>>>>>>> parameter that, if to small, won't mitigate the problem, but
>>>>> if to
>>>>> >>>>>>>>> big, greatly increases latency. (Possibly one could define a
>>>>> dynamic
>>>>> >>>>>>>>> session-like window to solve this though...) It also might
>>>>> be harder
>>>>> >>>>>>>>> for runners that *can* cheaply present stuff in timestamp
>>>>> order to
>>>>> >>>>>>>>> optimize. (That and, in practice, our annotation-style
>>>>> process methods
>>>>> >>>>>>>>> don't lend themselves to easy composition.) I think it could
>>>>> work in
>>>>> >>>>>>>>> specific cases though.
>>>>> >>>>>>>>>
>>>>> >>>>>>>>> More inline below.
>>>>> >>>>>>>>>
>>>>> >>>>>>>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <
>>>>> je.ik@seznam.cz> wrote:
>>>>> >>>>>>>>>> Hi Robert,
>>>>> >>>>>>>>>>
>>>>> >>>>>>>>>> > Beam has an exactly-once model. If the data was
>>>>> consumed, state
>>>>> >>>>>>>>>> mutated, and outputs written downstream (these three are
>>>>> committed
>>>>> >>>>>>>>>> together atomically) it will not be replayed. That does
>>>>> not, of
>>>>> >>>>>>>>>> course,
>>>>> >>>>>>>>>> solve the non-determanism due to ordering (including the
>>>>> fact that two
>>>>> >>>>>>>>>> operations reading the same PCollection may view different
>>>>> ordering).
>>>>> >>>>>>>>>>
>>>>> >>>>>>>>>> I think what you describe is a property of a runner, not of
>>>>> the model,
>>>>> >>>>>>>>>> right? I think if I run my pipeline on Flink I will not get
>>>>> this
>>>>> >>>>>>>>>> atomicity, because although Flink uses also exactly-once
>>>>> model if
>>>>> >>>>>>>>>> might
>>>>> >>>>>>>>>> write outputs multiple times.
>>>>> >>>>>>>>> Actually, I think it is a larger (open) question whether
>>>>> exactly once
>>>>> >>>>>>>>> is guaranteed by the model or whether runners are allowed to
>>>>> relax
>>>>> >>>>>>>>> that. I would think, however, that sources correctly
>>>>> implemented
>>>>> >>>>>>>>> should be idempotent when run atop an exactly once
>>>>> infrastructure such
>>>>> >>>>>>>>> as Flink of Dataflow.
>>>>> >>>>>>>>>
>>>>> >>>>>>>>>> > 1) Is it correct for a (Stateful)DoFn to assume
>>>>> elements are
>>>>> >>>>>>>>>> received
>>>>> >>>>>>>>>> in a specific order? In the current model, it is not. Being
>>>>> able to
>>>>> >>>>>>>>>> read, handle, and produced out-of-order data, including
>>>>> late data,
>>>>> >>>>>>>>>> is a
>>>>> >>>>>>>>>> pretty fundamental property of distributed systems.
>>>>> >>>>>>>>>>
>>>>> >>>>>>>>>> Yes, absolutely. The argument here is not that Stateful
>>>>> ParDo should
>>>>> >>>>>>>>>> presume to receive elements in any order, but to _present_
>>>>> it as
>>>>> >>>>>>>>>> such to
>>>>> >>>>>>>>>> the user @ProcessElement function.
>>>>> >>>>>>>>> Sounds like we should make this clearer.
>>>>> >>>>>>>>>
>>>>> >>>>>>>>>> > 2) Given that some operations are easier (or possibly
>>>>> only
>>>>> >>>>>>>>>> possible)
>>>>> >>>>>>>>>> to write when operating on ordered data, and that different
>>>>> runners
>>>>> >>>>>>>>>> may
>>>>> >>>>>>>>>> have (significantly) cheaper ways to provide this ordering
>>>>> than can be
>>>>> >>>>>>>>>> done by the user themselves, should we elevate this to a
>>>>> property of
>>>>> >>>>>>>>>> (Stateful?)DoFns that the runner can provide? I think a
>>>>> compelling
>>>>> >>>>>>>>>> argument can be made here that we should.
>>>>> >>>>>>>>>>
>>>>> >>>>>>>>>> +1
>>>>> >>>>>>>>>>
>>>>> >>>>>>>>>> Jan
>>>>> >>>>>>>>>>
>>>>> >>>>>>>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>>>>> >>>>>>>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <
>>>>> je.ik@seznam.cz> wrote:
>>>>> >>>>>>>>>>>> > I don't see batch vs. streaming as part of the
>>>>> model. One
>>>>> >>>>>>>>>>>> can have
>>>>> >>>>>>>>>>>> microbatch, or even a runner that alternates between
>>>>> different
>>>>> >>>>>>>>>>>> modes.
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> Although I understand motivation of this statement, this
>>>>> project
>>>>> >>>>>>>>>>>> name is
>>>>> >>>>>>>>>>>> "Apache Beam: An advanced unified programming model".
>>>>> What does the
>>>>> >>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the
>>>>> model?
>>>>> >>>>>>>>>>> What I mean is that streaming vs. batch is no longer part
>>>>> of the
>>>>> >>>>>>>>>>> model
>>>>> >>>>>>>>>>> (or ideally API), but pushed down to be a concern of the
>>>>> runner
>>>>> >>>>>>>>>>> (executor) of the pipeline.
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <
>>>>> je.ik@seznam.cz>
>>>>> >>>>>>>>>>> wrote:
>>>>> >>>>>>>>>>>> Hi Kenn,
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> OK, so if we introduce annotation, we can have stateful
>>>>> ParDo
>>>>> >>>>>>>>>>>> with sorting, that would perfectly resolve my issues. I
>>>>> still
>>>>> >>>>>>>>>>>> have some doubts, though. Let me explain. The current
>>>>> behavior of
>>>>> >>>>>>>>>>>> stateful ParDo has the following properties:
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> a) might fail in batch, although runs fine in
>>>>> streaming (that
>>>>> >>>>>>>>>>>> is due to the buffering, and unbounded lateness in batch,
>>>>> which
>>>>> >>>>>>>>>>>> was discussed back and forth in this thread)
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> b) might be non deterministic (this is because the
>>>>> elements
>>>>> >>>>>>>>>>>> arrive at somewhat random order, and even if you do the
>>>>> operation
>>>>> >>>>>>>>>>>> "assign unique ID to elements" this might produce
>>>>> different
>>>>> >>>>>>>>>>>> results when run multiple times)
>>>>> >>>>>>>>>>> PCollections are *explicitly* unordered. Any operations
>>>>> that
>>>>> >>>>>>>>>>> assume or
>>>>> >>>>>>>>>>> depend on a specific ordering for correctness (or
>>>>> determinism) must
>>>>> >>>>>>>>>>> provide that ordering themselves (i.e. tolerate "arbitrary
>>>>> shuffling
>>>>> >>>>>>>>>>> of inputs"). As you point out, that may be very expensive
>>>>> if you have
>>>>> >>>>>>>>>>> very hot keys with very large (unbounded) timestamp skew.
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>> StatefulDoFns are low-level operations that should be used
>>>>> with care;
>>>>> >>>>>>>>>>> the simpler windowing model gives determinism in the face
>>>>> of
>>>>> >>>>>>>>>>> unordered
>>>>> >>>>>>>>>>> data (though late data and non-end-of-window triggering
>>>>> introduces
>>>>> >>>>>>>>>>> some of the non-determanism back in).
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>>> What worries me most is the property b), because it seems
>>>>> to me
>>>>> >>>>>>>>>>>> to have serious consequences - not only that if you run
>>>>> twice
>>>>> >>>>>>>>>>>> batch pipeline you would get different results, but even
>>>>> on
>>>>> >>>>>>>>>>>> streaming, when pipeline fails and gets restarted from
>>>>> >>>>>>>>>>>> checkpoint, produced output might differ from the
>>>>> previous run
>>>>> >>>>>>>>>>>> and data from the first run might have already been
>>>>> persisted
>>>>> >>>>>>>>>>>> into sink. That would create somewhat messy outputs.
>>>>> >>>>>>>>>>> Beam has an exactly-once model. If the data was consumed,
>>>>> state
>>>>> >>>>>>>>>>> mutated, and outputs written downstream (these three are
>>>>> committed
>>>>> >>>>>>>>>>> together atomically) it will not be replayed. That does
>>>>> not, of
>>>>> >>>>>>>>>>> course, solve the non-determanism due to ordering
>>>>> (including the fact
>>>>> >>>>>>>>>>> that two operations reading the same PCollection may view
>>>>> different
>>>>> >>>>>>>>>>> ordering).
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>>> These two properties makes me think that the current
>>>>> >>>>>>>>>>>> implementation is more of a _special case_ than the
>>>>> general one.
>>>>> >>>>>>>>>>>> The general one would be that your state doesn't have the
>>>>> >>>>>>>>>>>> properties to be able to tolerate buffering problems
>>>>> and/or
>>>>> >>>>>>>>>>>> non-determinism. Which is the case where you need sorting
>>>>> in both
>>>>> >>>>>>>>>>>> streaming and batch to be part of the model.
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> Let me point out one more analogy - that is merging vs.
>>>>> >>>>>>>>>>>> non-merging windows. The general case (merging windows)
>>>>> implies
>>>>> >>>>>>>>>>>> sorting by timestamp in both batch case (explicit) and
>>>>> streaming
>>>>> >>>>>>>>>>>> (buffering). The special case (non-merging windows)
>>>>> doesn't rely
>>>>> >>>>>>>>>>>> on any timestamp ordering, so the sorting and buffering
>>>>> can be
>>>>> >>>>>>>>>>>> dropped. The underlying root cause of this is the same
>>>>> for both
>>>>> >>>>>>>>>>>> stateful ParDo and windowing (essentially, assigning
>>>>> window
>>>>> >>>>>>>>>>>> labels is a stateful operation when windowing function is
>>>>> merging).
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> The reason for the current behavior of stateful ParDo
>>>>> seems to be
>>>>> >>>>>>>>>>>> performance, but is it right to abandon correctness in
>>>>> favor of
>>>>> >>>>>>>>>>>> performance? Wouldn't it be more consistent to have the
>>>>> default
>>>>> >>>>>>>>>>>> behavior prefer correctness and when you have the specific
>>>>> >>>>>>>>>>>> conditions of state function having special properties,
>>>>> then you
>>>>> >>>>>>>>>>>> can annotate your DoFn (with something like
>>>>> >>>>>>>>>>>> @TimeOrderingAgnostic), which would yield a better
>>>>> performance in
>>>>> >>>>>>>>>>>> that case?
>>>>> >>>>>>>>>>> There are two separable questions here.
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>> 1) Is it correct for a (Stateful)DoFn to assume elements
>>>>> are received
>>>>> >>>>>>>>>>> in a specific order? In the current model, it is not.
>>>>> Being able to
>>>>> >>>>>>>>>>> read, handle, and produced out-of-order data, including
>>>>> late data, is
>>>>> >>>>>>>>>>> a pretty fundamental property of distributed systems.
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>> 2) Given that some operations are easier (or possibly only
>>>>> possible)
>>>>> >>>>>>>>>>> to write when operating on ordered data, and that
>>>>> different runners
>>>>> >>>>>>>>>>> may have (significantly) cheaper ways to provide this
>>>>> ordering than
>>>>> >>>>>>>>>>> can be done by the user themselves, should we elevate this
>>>>> to a
>>>>> >>>>>>>>>>> property of (Stateful?)DoFns that the runner can provide?
>>>>> I think a
>>>>> >>>>>>>>>>> compelling argument can be made here that we should.
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>> - Robert
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>>
>>>>> >>>>>>>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> Thanks for the nice small example of a calculation that
>>>>> depends
>>>>> >>>>>>>>>>>> on order. You are right that many state machines have this
>>>>> >>>>>>>>>>>> property. I agree w/ you and Luke that it is convenient
>>>>> for batch
>>>>> >>>>>>>>>>>> processing to sort by event timestamp before running a
>>>>> stateful
>>>>> >>>>>>>>>>>> ParDo. In streaming you could also implement "sort by
>>>>> event
>>>>> >>>>>>>>>>>> timestamp" by buffering until you know all earlier data
>>>>> will be
>>>>> >>>>>>>>>>>> dropped - a slack buffer up to allowed lateness.
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> I do not think that it is OK to sort in batch and not in
>>>>> >>>>>>>>>>>> streaming. Many state machines diverge very rapidly when
>>>>> things
>>>>> >>>>>>>>>>>> are out of order. So each runner if they see the
>>>>> >>>>>>>>>>>> "@OrderByTimestamp" annotation (or whatever) needs to
>>>>> deliver
>>>>> >>>>>>>>>>>> sorted data (by some mix of buffering and dropping), or
>>>>> to reject
>>>>> >>>>>>>>>>>> the pipeline as unsupported.
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> And also want to say that this is not the default case -
>>>>> many
>>>>> >>>>>>>>>>>> uses of state & timers in ParDo yield different results
>>>>> at the
>>>>> >>>>>>>>>>>> element level, but the results are equivalent at in the
>>>>> big
>>>>> >>>>>>>>>>>> picture. Such as the example of "assign a unique sequence
>>>>> number
>>>>> >>>>>>>>>>>> to each element" or "group into batches" it doesn't matter
>>>>> >>>>>>>>>>>> exactly what the result is, only that it meets the spec.
>>>>> And
>>>>> >>>>>>>>>>>> other cases like user funnels are monotonic enough that
>>>>> you also
>>>>> >>>>>>>>>>>> don't actually need sorting.
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> Kenn
>>>>> >>>>>>>>>>>>
>>>>> >>>>>>>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <
>>>>> je.ik@seznam.cz>
>>>>> >>>>>>>>>>>> wrote:
>>>>> >>>>>>>>>>>>> Yes, the problem will arise probably mostly when you
>>>>> have not
>>>>> >>>>>>>>>>>>> well distributed keys (or too few keys). I'm really not
>>>>> sure if
>>>>> >>>>>>>>>>>>> a pure GBK with a trigger can solve this - it might help
>>>>> to have
>>>>> >>>>>>>>>>>>> data driven trigger. There would still be some doubts,
>>>>> though.
>>>>> >>>>>>>>>>>>> The main question is still here - people say, that
>>>>> sorting by
>>>>> >>>>>>>>>>>>> timestamp before stateful ParDo would be prohibitively
>>>>> slow, but
>>>>> >>>>>>>>>>>>> I don't really see why - the sorting is very probably
>>>>> already
>>>>> >>>>>>>>>>>>> there. And if not (hash grouping instead of sorted
>>>>> grouping),
>>>>> >>>>>>>>>>>>> then the sorting would affect only user defined
>>>>> StatefulParDos.
>>>>> >>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>> This would suggest that the best way out of this would
>>>>> be really
>>>>> >>>>>>>>>>>>> to add annotation, so that the author of the pipeline
>>>>> can decide.
>>>>> >>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>> If that would be acceptable I think I can try to prepare
>>>>> some
>>>>> >>>>>>>>>>>>> basic functionality, but I'm not sure, if I would be
>>>>> able to
>>>>> >>>>>>>>>>>>> cover all runners / sdks.
>>>>> >>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>>>>> >>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>> It is read all per key and window and not just read all
>>>>> (this
>>>>> >>>>>>>>>>>>> still won't scale with hot keys in the global window).
>>>>> The GBK
>>>>> >>>>>>>>>>>>> preceding the StatefulParDo will guarantee that you are
>>>>> >>>>>>>>>>>>> processing all the values for a specific key and window
>>>>> at any
>>>>> >>>>>>>>>>>>> given time. Is there a specific window/trigger that is
>>>>> missing
>>>>> >>>>>>>>>>>>> that you feel would remove the need for you to use
>>>>> StatefulParDo?
>>>>> >>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <
>>>>> je.ik@seznam.cz>
>>>>> >>>>>>>>>>>>> wrote:
>>>>> >>>>>>>>>>>>>> Hi Lukasz,
>>>>> >>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> Today, if you must have a strict order, you must
>>>>> guarantee
>>>>> >>>>>>>>>>>>>>> that your StatefulParDo implements the necessary
>>>>> "buffering &
>>>>> >>>>>>>>>>>>>>> sorting" into state.
>>>>> >>>>>>>>>>>>>> Yes, no problem with that. But this whole discussion
>>>>> started,
>>>>> >>>>>>>>>>>>>> because *this doesn't work on batch*. You simply cannot
>>>>> first
>>>>> >>>>>>>>>>>>>> read everything from distributed storage and then
>>>>> buffer it all
>>>>> >>>>>>>>>>>>>> into memory, just to read it again, but sorted. That
>>>>> will not
>>>>> >>>>>>>>>>>>>> work. And even if it would, it would be a terrible
>>>>> waste of
>>>>> >>>>>>>>>>>>>> resources.
>>>>> >>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>> Jan
>>>>> >>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>>>>> >>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <
>>>>> je.ik@seznam.cz>
>>>>> >>>>>>>>>>>>>> wrote:
>>>>> >>>>>>>>>>>>>>> This discussion brings many really interesting
>>>>> questions for
>>>>> >>>>>>>>>>>>>>> me. :-)
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> > I don't see batch vs. streaming as part of the
>>>>> model. One
>>>>> >>>>>>>>>>>>>>> can have
>>>>> >>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
>>>>> different
>>>>> >>>>>>>>>>>>>>> modes.
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> Although I understand motivation of this statement,
>>>>> this
>>>>> >>>>>>>>>>>>>>> project name is
>>>>> >>>>>>>>>>>>>>> "Apache Beam: An advanced unified programming model".
>>>>> What
>>>>> >>>>>>>>>>>>>>> does the
>>>>> >>>>>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of
>>>>> the model?
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> Using microbatching, chaining of batch jobs, or pure
>>>>> streaming
>>>>> >>>>>>>>>>>>>>> are
>>>>> >>>>>>>>>>>>>>> exactly the "runtime conditions/characteristics" I
>>>>> refer to.
>>>>> >>>>>>>>>>>>>>> All these
>>>>> >>>>>>>>>>>>>>> define several runtime parameters, which in turn
>>>>> define how
>>>>> >>>>>>>>>>>>>>> well/badly
>>>>> >>>>>>>>>>>>>>> will the pipeline perform and how many resources might
>>>>> be
>>>>> >>>>>>>>>>>>>>> needed. From
>>>>> >>>>>>>>>>>>>>> my point of view, pure streaming should be the most
>>>>> resource
>>>>> >>>>>>>>>>>>>>> demanding
>>>>> >>>>>>>>>>>>>>> (if not, why bother with batch? why not run everything
>>>>> in
>>>>> >>>>>>>>>>>>>>> streaming
>>>>> >>>>>>>>>>>>>>> only? what will there remain to "unify"?).
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> > Fortunately, for batch, only the state for a
>>>>> single key
>>>>> >>>>>>>>>>>>>>> needs to be
>>>>> >>>>>>>>>>>>>>> preserved at a time, rather than the state for all
>>>>> keys across
>>>>> >>>>>>>>>>>>>>> the range
>>>>> >>>>>>>>>>>>>>> of skew. Of course if you have few or hot keys, one
>>>>> can still
>>>>> >>>>>>>>>>>>>>> have
>>>>> >>>>>>>>>>>>>>> issues (and this is not specific to StatefulDoFns).
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> Yes, but here is still the presumption that my
>>>>> stateful DoFn can
>>>>> >>>>>>>>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain
>>>>> the use
>>>>> >>>>>>>>>>>>>>> case in
>>>>> >>>>>>>>>>>>>>> more detail.
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> Suppose you have input stream consisting of 1s and 0s
>>>>> (and
>>>>> >>>>>>>>>>>>>>> some key for
>>>>> >>>>>>>>>>>>>>> each element, which is irrelevant for the
>>>>> demonstration). Your
>>>>> >>>>>>>>>>>>>>> task is
>>>>> >>>>>>>>>>>>>>> to calculate in running global window the actual
>>>>> number of
>>>>> >>>>>>>>>>>>>>> changes
>>>>> >>>>>>>>>>>>>>> between state 0 and state 1 and vice versa. When the
>>>>> state
>>>>> >>>>>>>>>>>>>>> doesn't
>>>>> >>>>>>>>>>>>>>> change, you don't calculate anything. If input (for
>>>>> given key)
>>>>> >>>>>>>>>>>>>>> would be
>>>>> >>>>>>>>>>>>>>> (tN denotes timestamp N):
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t1: 1
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t2: 0
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t3: 0
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t4: 1
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t5: 1
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t6: 0
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> then the output should yield (supposing that default
>>>>> state is
>>>>> >>>>>>>>>>>>>>> zero):
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t1: (one: 1, zero: 0)
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t2: (one: 1, zero: 1)
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t3: (one: 1, zero: 1)
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t4: (one: 2, zero: 1)
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t5: (one: 2, zero: 1)
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> t6: (one: 2, zero: 2)
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> How would you implement this in current Beam semantics?
>>>>> >>>>>>>>>>>>>> I think your saying here that I know that my input is
>>>>> ordered
>>>>> >>>>>>>>>>>>>> in a specific way and since I assume the order when
>>>>> writing my
>>>>> >>>>>>>>>>>>>> pipeline I can perform this optimization. But there is
>>>>> nothing
>>>>> >>>>>>>>>>>>>> preventing a runner from noticing that your processing
>>>>> in the
>>>>> >>>>>>>>>>>>>> global window with a specific type of trigger and
>>>>> re-ordering
>>>>> >>>>>>>>>>>>>> your inputs/processing to get better performance (since
>>>>> you
>>>>> >>>>>>>>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>>>>> >>>>>>>>>>>>>> streaming for the GlobalWindow).
>>>>> >>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>> Today, if you must have a strict order, you must
>>>>> guarantee that
>>>>> >>>>>>>>>>>>>> your StatefulParDo implements the necessary "buffering &
>>>>> >>>>>>>>>>>>>> sorting" into state. I can see why you would want an
>>>>> annotation
>>>>> >>>>>>>>>>>>>> that says I must have timestamp ordered elements, since
>>>>> it
>>>>> >>>>>>>>>>>>>> makes writing certain StatefulParDos much easier.
>>>>> StatefulParDo
>>>>> >>>>>>>>>>>>>> is a low-level function, it really is the "here you go
>>>>> and do
>>>>> >>>>>>>>>>>>>> whatever you need to but here be dragons" function while
>>>>> >>>>>>>>>>>>>> windowing and triggering is meant to keep many people
>>>>> from
>>>>> >>>>>>>>>>>>>> writing StatefulParDo in the first place.
>>>>> >>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> > Pipelines that fail in the "worst case" batch
>>>>> scenario
>>>>> >>>>>>>>>>>>>>> are likely to
>>>>> >>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the
>>>>> watermark
>>>>> >>>>>>>>>>>>>>> falls
>>>>> >>>>>>>>>>>>>>> behind in streaming mode as well.
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> But the worst case is defined by input of size
>>>>> (available
>>>>> >>>>>>>>>>>>>>> resources +
>>>>> >>>>>>>>>>>>>>> single byte) -> pipeline fail. Although it could have
>>>>> >>>>>>>>>>>>>>> finished, given
>>>>> >>>>>>>>>>>>>>> the right conditions.
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> > This might be reasonable, implemented by default
>>>>> by
>>>>> >>>>>>>>>>>>>>> buffering
>>>>> >>>>>>>>>>>>>>> everything and releasing elements as the watermark
>>>>> (+lateness)
>>>>> >>>>>>>>>>>>>>> advances,
>>>>> >>>>>>>>>>>>>>> but would likely lead to inefficient (though *maybe*
>>>>> easier to
>>>>> >>>>>>>>>>>>>>> reason
>>>>> >>>>>>>>>>>>>>> about) code.
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> Sure, the pipeline will be less efficient, because it
>>>>> would
>>>>> >>>>>>>>>>>>>>> have to
>>>>> >>>>>>>>>>>>>>> buffer and sort the inputs. But at least it will
>>>>> produce
>>>>> >>>>>>>>>>>>>>> correct results
>>>>> >>>>>>>>>>>>>>> in cases where updates to state are order-sensitive.
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> > Would it be roughly equivalent to GBK +
>>>>> FlatMap(lambda
>>>>> >>>>>>>>>>>>>>> (key, values):
>>>>> >>>>>>>>>>>>>>> [(key, value) for value in values])?
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> I'd say roughly yes, but difference would be in the
>>>>> trigger.
>>>>> >>>>>>>>>>>>>>> The trigger
>>>>> >>>>>>>>>>>>>>> should ideally fire as soon as watermark (+lateness)
>>>>> crosses
>>>>> >>>>>>>>>>>>>>> element
>>>>> >>>>>>>>>>>>>>> with lowest timestamp in the buffer. Although this
>>>>> could be
>>>>> >>>>>>>>>>>>>>> somehow
>>>>> >>>>>>>>>>>>>>> emulated by fixed trigger each X millis.
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> > Or is the underlying desire just to be able to
>>>>> hint to
>>>>> >>>>>>>>>>>>>>> the runner
>>>>> >>>>>>>>>>>>>>> that the code may perform better (e.g. require less
>>>>> resources)
>>>>> >>>>>>>>>>>>>>> as skew
>>>>> >>>>>>>>>>>>>>> is reduced (and hence to order by timestamp iff it's
>>>>> cheap)?
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> No, the sorting would have to be done in streaming
>>>>> case as
>>>>> >>>>>>>>>>>>>>> well. That is
>>>>> >>>>>>>>>>>>>>> an imperative of the unified model. I think it is
>>>>> possible to
>>>>> >>>>>>>>>>>>>>> sort by
>>>>> >>>>>>>>>>>>>>> timestamp only in batch case (and do it for *all* batch
>>>>> >>>>>>>>>>>>>>> stateful pardos
>>>>> >>>>>>>>>>>>>>> without annotation), or introduce annotation, but then
>>>>> make
>>>>> >>>>>>>>>>>>>>> the same
>>>>> >>>>>>>>>>>>>>> guarantees for streaming case as well.
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> Jan
>>>>> >>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>>>>> >>>>>>>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>>>>> >>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>>>>> >>>>>>>>>>>>>>>>> Hi Robert,
>>>>> >>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>> yes, I think you rephrased my point - although no
>>>>> *explicit*
>>>>> >>>>>>>>>>>>>>>>> guarantees
>>>>> >>>>>>>>>>>>>>>>> of ordering are given in either mode, there is
>>>>> *implicit*
>>>>> >>>>>>>>>>>>>>>>> ordering in
>>>>> >>>>>>>>>>>>>>>>> streaming case that is due to nature of the
>>>>> processing - the
>>>>> >>>>>>>>>>>>>>>>> difference
>>>>> >>>>>>>>>>>>>>>>> between watermark and timestamp of elements flowing
>>>>> through
>>>>> >>>>>>>>>>>>>>>>> the pipeline
>>>>> >>>>>>>>>>>>>>>>> are generally low (too high difference leads to the
>>>>> >>>>>>>>>>>>>>>>> overbuffering
>>>>> >>>>>>>>>>>>>>>>> problem), but there is no such bound on batch.
>>>>> >>>>>>>>>>>>>>>> Fortunately, for batch, only the state for a single
>>>>> key needs
>>>>> >>>>>>>>>>>>>>>> to be
>>>>> >>>>>>>>>>>>>>>> preserved at a time, rather than the state for all
>>>>> keys
>>>>> >>>>>>>>>>>>>>>> across the
>>>>> >>>>>>>>>>>>>>>> range of skew. Of course if you have few or hot keys,
>>>>> one can
>>>>> >>>>>>>>>>>>>>>> still
>>>>> >>>>>>>>>>>>>>>> have issues (and this is not specific to
>>>>> StatefulDoFns).
>>>>> >>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>> As a result, I see a few possible solutions:
>>>>> >>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>> - the best and most natural seems to be
>>>>> extension of
>>>>> >>>>>>>>>>>>>>>>> the model, so
>>>>> >>>>>>>>>>>>>>>>> that it defines batch as not only "streaming pipeline
>>>>> >>>>>>>>>>>>>>>>> executed in batch
>>>>> >>>>>>>>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>>>>> >>>>>>>>>>>>>>>>> characteristics as
>>>>> >>>>>>>>>>>>>>>>> in streaming case, executed in batch fashion", I
>>>>> really
>>>>> >>>>>>>>>>>>>>>>> don't think that
>>>>> >>>>>>>>>>>>>>>>> there are any conflicts with the current model, or
>>>>> that this
>>>>> >>>>>>>>>>>>>>>>> could
>>>>> >>>>>>>>>>>>>>>>> affect performance, because the required sorting (as
>>>>> pointed by
>>>>> >>>>>>>>>>>>>>>>> Aljoscha) is very probably already done during
>>>>> translation
>>>>> >>>>>>>>>>>>>>>>> of stateful
>>>>> >>>>>>>>>>>>>>>>> pardos. Also note that this definition only affects
>>>>> user
>>>>> >>>>>>>>>>>>>>>>> defined
>>>>> >>>>>>>>>>>>>>>>> stateful pardos
>>>>> >>>>>>>>>>>>>>>> I don't see batch vs. streaming as part of the model.
>>>>> One can
>>>>> >>>>>>>>>>>>>>>> have
>>>>> >>>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
>>>>> >>>>>>>>>>>>>>>> different modes.
>>>>> >>>>>>>>>>>>>>>> The model describes what the valid outputs are given a
>>>>> >>>>>>>>>>>>>>>> (sometimes
>>>>> >>>>>>>>>>>>>>>> partial) set of inputs. It becomes really hard to
>>>>> define
>>>>> >>>>>>>>>>>>>>>> things like
>>>>> >>>>>>>>>>>>>>>> "as good runtime characteristics." Once you allow any
>>>>> >>>>>>>>>>>>>>>> out-of-orderedness, it is not very feasible to try
>>>>> and define
>>>>> >>>>>>>>>>>>>>>> (and
>>>>> >>>>>>>>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>>>>> >>>>>>>>>>>>>>>> out-of-orderedness.
>>>>> >>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>> Pipelines that fail in the "worst case" batch
>>>>> scenario are
>>>>> >>>>>>>>>>>>>>>> likely to
>>>>> >>>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the
>>>>> watermark
>>>>> >>>>>>>>>>>>>>>> falls
>>>>> >>>>>>>>>>>>>>>> behind in streaming mode as well.
>>>>> >>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>> - another option would be to introduce
>>>>> annotation for
>>>>> >>>>>>>>>>>>>>>>> DoFns (e.g.
>>>>> >>>>>>>>>>>>>>>>> @RequiresStableTimeCharacteristics), which would
>>>>> result in
>>>>> >>>>>>>>>>>>>>>>> the sorting
>>>>> >>>>>>>>>>>>>>>>> in batch case - but - this extension would have to
>>>>> ensure
>>>>> >>>>>>>>>>>>>>>>> the sorting in
>>>>> >>>>>>>>>>>>>>>>> streaming mode also - it would require definition of
>>>>> allowed
>>>>> >>>>>>>>>>>>>>>>> lateness,
>>>>> >>>>>>>>>>>>>>>>> and triggger (essentially similar to window)
>>>>> >>>>>>>>>>>>>>>> This might be reasonable, implemented by default by
>>>>> buffering
>>>>> >>>>>>>>>>>>>>>> everything and releasing elements as the watermark
>>>>> (+lateness)
>>>>> >>>>>>>>>>>>>>>> advances, but would likely lead to inefficient (though
>>>>> >>>>>>>>>>>>>>>> *maybe* easier
>>>>> >>>>>>>>>>>>>>>> to reason about) code. Not sure about the semantics of
>>>>> >>>>>>>>>>>>>>>> triggering
>>>>> >>>>>>>>>>>>>>>> here, especially data-driven triggers. Would it be
>>>>> roughly
>>>>> >>>>>>>>>>>>>>>> equivalent
>>>>> >>>>>>>>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value)
>>>>> for
>>>>> >>>>>>>>>>>>>>>> value in
>>>>> >>>>>>>>>>>>>>>> values])?
>>>>> >>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>> Or is the underlying desire just to be able to hint
>>>>> to the
>>>>> >>>>>>>>>>>>>>>> runner that
>>>>> >>>>>>>>>>>>>>>> the code may perform better (e.g. require less
>>>>> resources) as
>>>>> >>>>>>>>>>>>>>>> skew is
>>>>> >>>>>>>>>>>>>>>> reduced (and hence to order by timestamp iff it's
>>>>> cheap)?
>>>>> >>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>> - last option would be to introduce these
>>>>> "higher order
>>>>> >>>>>>>>>>>>>>>>> guarantees" in
>>>>> >>>>>>>>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems
>>>>> to be the
>>>>> >>>>>>>>>>>>>>>>> worst
>>>>> >>>>>>>>>>>>>>>>> option to me
>>>>> >>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>> I see the first two options quite equally good,
>>>>> although the
>>>>> >>>>>>>>>>>>>>>>> letter one
>>>>> >>>>>>>>>>>>>>>>> is probably more time consuming to implement. But it
>>>>> would
>>>>> >>>>>>>>>>>>>>>>> bring
>>>>> >>>>>>>>>>>>>>>>> additional feature to streaming case as well.
>>>>> >>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>> Thanks for any thoughts.
>>>>> >>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>> Jan
>>>>> >>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>>>>> >>>>>>>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>>>>> >>>>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>>>>> >>>>>>>>>>>>>>>>>>> Hi Reuven,
>>>>> >>>>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in
>>>>> batch
>>>>> >>>>>>>>>>>>>>>>>>>> runners.
>>>>> >>>>>>>>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic
>>>>> inside
>>>>> >>>>>>>>>>>>>>>>>>> the state works for absolutely unbounded
>>>>> out-of-orderness
>>>>> >>>>>>>>>>>>>>>>>>> of elements. That basically (practically) can work
>>>>> only
>>>>> >>>>>>>>>>>>>>>>>>> for cases, where the order of input elements
>>>>> doesn't
>>>>> >>>>>>>>>>>>>>>>>>> matter. But, "state" can refer to "state machine",
>>>>> and any
>>>>> >>>>>>>>>>>>>>>>>>> time you have a state machine involved, then the
>>>>> ordering
>>>>> >>>>>>>>>>>>>>>>>>> of elements would matter.
>>>>> >>>>>>>>>>>>>>>>>> No guarantees on order are provided in *either*
>>>>> streaming
>>>>> >>>>>>>>>>>>>>>>>> or batch
>>>>> >>>>>>>>>>>>>>>>>> mode by the model. However, it is the case that in
>>>>> order to
>>>>> >>>>>>>>>>>>>>>>>> make
>>>>> >>>>>>>>>>>>>>>>>> forward progress most streaming runners attempt to
>>>>> limit
>>>>> >>>>>>>>>>>>>>>>>> the amount of
>>>>> >>>>>>>>>>>>>>>>>> out-of-orderedness of elements (in terms of event
>>>>> time vs.
>>>>> >>>>>>>>>>>>>>>>>> processing
>>>>> >>>>>>>>>>>>>>>>>> time) to make forward progress, which in turn could
>>>>> help
>>>>> >>>>>>>>>>>>>>>>>> cap the
>>>>> >>>>>>>>>>>>>>>>>> amount of state that must be held concurrently,
>>>>> whereas a
>>>>> >>>>>>>>>>>>>>>>>> batch runner
>>>>> >>>>>>>>>>>>>>>>>> may not allow any state to be safely discarded
>>>>> until the whole
>>>>> >>>>>>>>>>>>>>>>>> timeline from infinite past to infinite future has
>>>>> been
>>>>> >>>>>>>>>>>>>>>>>> observed.
>>>>> >>>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>>> Also, as pointed out, state is not preserved "batch
>>>>> to
>>>>> >>>>>>>>>>>>>>>>>> batch" in batch mode.
>>>>> >>>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>>>>> >>>>>>>>>>>>>>>>>> <mx...@apache.org> wrote:
>>>>> >>>>>>>>>>>>>>>>>>
>>>>> >>>>>>>>>>>>>>>>>>>> batch semantics and streaming semantics
>>>>> differs only
>>>>> >>>>>>>>>>>>>>>>>>>> in that I can have GlobalWindow with default
>>>>> trigger on
>>>>> >>>>>>>>>>>>>>>>>>>> batch and cannot on stream
>>>>> >>>>>>>>>>>>>>>>>>> You can have a GlobalWindow in streaming with a
>>>>> default
>>>>> >>>>>>>>>>>>>>>>>>> trigger. You
>>>>> >>>>>>>>>>>>>>>>>>> could define additional triggers that do early
>>>>> firings.
>>>>> >>>>>>>>>>>>>>>>>>> And you could
>>>>> >>>>>>>>>>>>>>>>>>> even trigger the global window by advancing the
>>>>> watermark
>>>>> >>>>>>>>>>>>>>>>>>> to +inf.
>>>>> >>>>>>>>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global
>>>>> window with
>>>>> >>>>>>>>>>>>>>>>>> default
>>>>> >>>>>>>>>>>>>>>>>> trigger on unbounded PCollections in the SDK
>>>>> because this
>>>>> >>>>>>>>>>>>>>>>>> is more
>>>>> >>>>>>>>>>>>>>>>>> likely to be user error than an actual desire to
>>>>> have no
>>>>> >>>>>>>>>>>>>>>>>> output until
>>>>> >>>>>>>>>>>>>>>>>> drain. But it's semantically valid in the model.
>>>>>
>>>>>
Re: Definition of Unified model
Posted by Lukasz Cwik <lc...@google.com>.
You don't need the ordering, you need support for higher dimensionality in
the element tracking so that a runner can track that it is making forward
progress, schedule and execute timers, know when side inputs are ready, ...
On Thu, May 30, 2019 at 4:06 AM Jan Lukavský <je...@seznam.cz> wrote:
> Hi Lukasz, do you suggest, that ordering can have some relation to
> iterations? That sounds interesting, and from the top of my head, it can be
> true - you have to order on iteration number to get consistent results.
> Maybe we are tracking something deeper here! :-)
>
> Jan
> On 5/29/19 7:02 PM, Lukasz Cwik wrote:
>
> Expanding the dimensionality could be the basis for loops within the graph
> since loops could be modeled as (time, loop iteration #, nested loop
> iteration #, nested nested loop iteration #, ...)
>
> On Tue, May 28, 2019 at 12:10 PM Jan Lukavský <je...@seznam.cz> wrote:
>
>> Could this be solved by "expanding the dimensionality" of time field?
>> What I mean by that - if input element to to FlatMap has sequence number T,
>> then the (stateless) FlatMap knows the ordering of output elements, right?
>> If it would expand the field by which it will next sort the elements to (X,
>> 1), (X, 2), ... (X, N), then it would be possible to sort the elements back
>> later. There seems to be no need for state to achieve that, or?
>>
>> Jan
>> On 5/28/19 6:52 PM, Reuven Lax wrote:
>>
>> A slightly larger concern: it also will force users to create stateful
>> DoFns everywhere to generate these sequence numbers. If I have a ParDo that
>> is not a simple 1:1 transform (i.e. not MapElements), then the ParDo will
>> need to generate its own sequence numbers for ordering, and the only safe
>> way to do so is to use a stateful DoFn. This turns what used to be a simple
>> in-memory DoFn into one that has to access state. Also I believe many
>> runners will not fuse stateful DoFns. While none of this poses a problem
>> for the model, it could make ordering extremely expensive to achieve.
>>
>> Reuven
>>
>> On Tue, May 28, 2019 at 6:09 AM Jan Lukavský <je...@seznam.cz> wrote:
>>
>>> Hi Reuven,
>>>
>>> > It also gets awkward with Flatten - the sequence number is no longer
>>> enough, you must also encode which side of the flatten each element came
>>> from.
>>>
>>> That is a generic need. Even if you read data from Kafka, the offsets
>>> are comparable only inside single partition. So, for Kafka to work as a
>>> FIFO for ordering, elements with same key have to be pushed to the same
>>> partition (otherwise Kafka cannot act as FIFO, because different partitions
>>> can be handled by different brokers, which means different observers and
>>> they therefore might not agree on the order of events). So if we want to
>>> emulate FIFO per key, then the sequence IDs have also be per key.
>>> On 5/28/19 2:33 PM, Reuven Lax wrote:
>>>
>>> Sequence metadata does have the disadvantage that users can no longer
>>> use the types coming from the source. You must create a new type that
>>> contains a sequence number (unless Beam provides this). It also gets
>>> awkward with Flatten - the sequence number is no longer enough, you must
>>> also encode which side of the flatten each element came from.
>>>
>>> On Tue, May 28, 2019 at 3:18 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>
>>>> As I understood it, Kenn was supporting the idea that sequence metadata
>>>> is preferable over FIFO. I was trying to point out, that it even should
>>>> provide the same functionally as FIFO, plus one important more -
>>>> reproducibility and ability to being persisted and reused the same way
>>>> in batch and streaming.
>>>>
>>>> There is no doubt, that sequence metadata can be stored in every
>>>> storage. But, regarding some implicit ordering that sources might have
>>>> -
>>>> yes, of course, data written into HDFS or Cloud Storage has ordering,
>>>> but only partial - inside some bulk (e.g. file) and the ordering is not
>>>> defined correctly on boundaries of these bulks (between files). That is
>>>> why I'd say, that ordering of sources is relevant only for
>>>> (partitioned!) streaming sources and generally always reduces to
>>>> sequence metadata (e.g. offsets).
>>>>
>>>> Jan
>>>>
>>>> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
>>>> > Huge +1 to all Kenn said.
>>>> >
>>>> > Jan, batch sources can have orderings too, just like Kafka. I think
>>>> > it's reasonable (for both batch and streaming) that if a source has an
>>>> > ordering that is an important part of the data, it should preserve
>>>> > this ordering into the data itself (e.g. as sequence numbers, offsets,
>>>> > etc.)
>>>> >
>>>> > On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org>
>>>> wrote:
>>>> >> I strongly prefer explicit sequence metadata over FIFO requirements,
>>>> because:
>>>> >>
>>>> >> - FIFO is complex to specify: for example Dataflow has "per stage
>>>> key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage"
>>>> is not a portable concept, nor even guaranteed to remain a Dataflow concept)
>>>> >> - complex specifications are by definition poor usability (if
>>>> necessary, then it is what it is)
>>>> >> - overly restricts the runner, reduces parallelism, for example
>>>> any non-stateful ParDo has per-element parallelism, not per "key"
>>>> >> - another perspective on that: FIFO makes everyone pay rather than
>>>> just the transform that requires exactly sequencing
>>>> >> - previous implementation details like reshuffles become part of
>>>> the model
>>>> >> - I'm not even convinced the use cases involved are addressed by
>>>> some careful FIFO restrictions; many sinks re-key and they would all have
>>>> to become aware of how keying of a sequence of "stages" affects the
>>>> end-to-end FIFO
>>>> >>
>>>> >> A noop becoming a non-noop is essentially the mathematical
>>>> definition of moving from higher-level to lower-level abstraction.
>>>> >>
>>>> >> So this strikes at the core question of what level of abstraction
>>>> Beam aims to represent. Lower-level means there are fewer possible
>>>> implementations and it is more tied to the underlying architecture, and
>>>> anything not near-exact match pays a huge penalty. Higher-level means there
>>>> are more implementations possible with different tradeoffs, though they may
>>>> all pay a minor penalty.
>>>> >>
>>>> >> I could be convinced to change my mind, but it needs some extensive
>>>> design, examples, etc. I think it is probably about the most consequential
>>>> design decision in the whole Beam model, around the same level as the
>>>> decision to use ParDo and GBK as the primitives IMO.
>>>> >>
>>>> >> Kenn
>>>> >>
>>>> >> On Thu, May 23, 2019 at 10:17 AM Reuven Lax <re...@google.com>
>>>> wrote:
>>>> >>> Not really. I'm suggesting that some variant of FIFO ordering is
>>>> necessary, which requires either runners natively support FIFO ordering or
>>>> transforms adding some extra sequence number to each record to sort by.
>>>> >>>
>>>> >>> I still think your proposal is very useful by the way. I'm merely
>>>> pointing out that to solve the state-machine problem we probably need
>>>> something more.
>>>> >>>
>>>> >>> Reuven
>>>> >>>
>>>> >>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je...@seznam.cz>
>>>> wrote:
>>>> >>>> Hi,
>>>> >>>> yes. It seems that ordering by user supplied UDF makes sense and I
>>>> will update the design proposal accordingly.
>>>> >>>> Would that solve the issues you mention?
>>>> >>>> Jan
>>>> >>>> ---------- Původní e-mail ----------
>>>> >>>> Od: Reuven Lax <re...@google.com>
>>>> >>>> Komu: dev <de...@beam.apache.org>
>>>> >>>> Datum: 23. 5. 2019 18:44:38
>>>> >>>> Předmět: Re: Definition of Unified model
>>>> >>>>
>>>> >>>> I'm simply saying that timestamp ordering is insufficient for
>>>> state machines. I wasn't proposing Kafka as a solution - that was simply an
>>>> example of how people solve this problem in other scenarios.
>>>> >>>>
>>>> >>>> BTW another example of ordering: Imagine today that you have a
>>>> triggered Sum aggregation writing out to a key-value sink. In theory we
>>>> provide no ordering, so the sink might write the triggered sums in the
>>>> wrong order, ending up with an incorrect value in the sink. In this case
>>>> you probably want values ordered by trigger pane index.
>>>> >>>>
>>>> >>>> Reuven
>>>> >>>>
>>>> >>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz>
>>>> wrote:
>>>> >>>>
>>>> >>>> Hi Reuven,
>>>> >>>> I share the view point of Robert. I think the isuue you refer to
>>>> is not in reality related to timestamps, but to the fact, that ordering of
>>>> events in time is observer dependent (either caused by relativity, or time
>>>> skew, essentially this has the same consequences). And the resolution in
>>>> fact isn't Kafka, but generally an authoritative observer, that tells you
>>>> "I saw the events in this order". And you either have one (and have the
>>>> outcome of his observation persisted in the data - e.g. as offset in Kafka
>>>> partition), then you should be able to use it (maybe that suggests afterall
>>>> that sorting by some user supplied UDF might make sense), or do not have
>>>> it, and then any interpretation of the data seems to be equally valid.
>>>> Although determinism is fine, of course.
>>>> >>>> Jan
>>>> >>>> ---------- Původní e-mail ----------
>>>> >>>> Od: Reuven Lax <re...@google.com>
>>>> >>>> Komu: dev <de...@beam.apache.org>
>>>> >>>> Datum: 23. 5. 2019 17:39:12
>>>> >>>> Předmět: Re: Definition of Unified model
>>>> >>>>
>>>> >>>> So an example would be elements of type "startUserSession" and
>>>> "endUserSession" (website sessions, not Beam sessions). Logically you may
>>>> need to process them in the correct order if you have any sort of
>>>> state-machine logic. However timestamp ordering is never guaranteed to
>>>> match the logical ordering. Not only might you have several elements with
>>>> the same timestamp, but in reality time skew across backend servers can
>>>> cause the events to have timestamps in reverse order of the actual
>>>> causality order.
>>>> >>>>
>>>> >>>> People do solve this problem today though. Publish the events to
>>>> Kafka, making sure that events for the same user end up in the same Kafka
>>>> partition. This ensures that the events appear in the Kafka partitions in
>>>> causality order, even if the timestamp order doesn't match. The your Kafka
>>>> subscriber simply process the elements in each partition in order.
>>>> >>>>
>>>> >>>> I think the ability to impose FIFO causality ordering is what's
>>>> needed for any state-machine work. Timestamp ordering has advantages
>>>> (though often I think the advantage is in state), but does not solve this
>>>> problem.
>>>> >>>>
>>>> >>>> Reuven
>>>> >>>>
>>>> >>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <
>>>> robertwb@google.com> wrote:
>>>> >>>>
>>>> >>>> Good point.
>>>> >>>>
>>>> >>>> The "implementation-specific" way I would do this is
>>>> >>>> window-by-instant, followed by a DoFn that gets all the elements
>>>> with
>>>> >>>> the same timestamp and sorts/acts accordingly, but this counts on
>>>> the
>>>> >>>> runner producing windows in timestamp order (likely?) and also the
>>>> >>>> subsequent DoFn getting them in this order (also likely, due to
>>>> >>>> fusion).
>>>> >>>>
>>>> >>>> One could make the argument that, though it does not provide
>>>> >>>> deterministic behavior, getting elements of the same timestamp in
>>>> >>>> different orders should produce equally valid interpretations of
>>>> the
>>>> >>>> data. (After all, due to relatively, timestamps are not technically
>>>> >>>> well ordered across space.) I can see how data-dependent
>>>> tiebreakers
>>>> >>>> could be useful, or promises of preservation of order between
>>>> >>>> operations.
>>>> >>>>
>>>> >>>> - Robert
>>>> >>>>
>>>> >>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com>
>>>> wrote:
>>>> >>>>> So Jan's example of state machines is quite a valid use case for
>>>> ordering. However in my experience, timestamp ordering is insufficient for
>>>> state machines. Elements that cause state transitions might come in with
>>>> the exact same timestamp, yet still have a necessary ordering. Especially
>>>> given Beam's decision to have milliseconds timestamps this is possible, but
>>>> even at microsecond or nanosecond precision this can happen at scale. To
>>>> handle state machines you usually need some sort of FIFO ordering along
>>>> with an ordered sources, such as Kafka, not timestamp ordering.
>>>> >>>>>
>>>> >>>>> Reuven
>>>> >>>>>
>>>> >>>>> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz>
>>>> wrote:
>>>> >>>>>> Hi all,
>>>> >>>>>>
>>>> >>>>>> thanks everyone for this discussion. I think I have gathered
>>>> enough
>>>> >>>>>> feedback to be able to put down a proposition for changes, which
>>>> I will
>>>> >>>>>> do and send to this list for further discussion. There are still
>>>> doubts
>>>> >>>>>> remaining the non-determinism and it's relation to outputs
>>>> stability vs.
>>>> >>>>>> latency. But I will try to clarify all this in the design
>>>> document.
>>>> >>>>>>
>>>> >>>>>> Thanks,
>>>> >>>>>>
>>>> >>>>>> Jan
>>>> >>>>>>
>>>> >>>>>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>>>> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
>>>> >>>>>>>> current understanding.
>>>> >>>>>>> In essence your description of how exactly-once works in Flink
>>>> is
>>>> >>>>>>> correct. The general assumption in Flink is that pipelines must
>>>> be
>>>> >>>>>>> deterministic and thus produce idempotent writes in the case of
>>>> >>>>>>> failures. However, that doesn't mean Beam sinks can't guarantee
>>>> a bit
>>>> >>>>>>> more with what Flink has to offer.
>>>> >>>>>>>
>>>> >>>>>>> Luke already mentioned the design discussions for
>>>> @RequiresStableInput
>>>> >>>>>>> which ensures idempotent writes for non-deterministic
>>>> pipelines. This
>>>> >>>>>>> is not part of the model but an optional Beam feature.
>>>> >>>>>>>
>>>> >>>>>>> We recently implemented support for @RequiresStableInput in the
>>>> Flink
>>>> >>>>>>> Runner. Reuven mentioned the Flink checkpoint confirmation,
>>>> which
>>>> >>>>>>> allows us to buffer (and checkpoint) processed data and only
>>>> emit it
>>>> >>>>>>> once a Flink checkpoint has completed.
>>>> >>>>>>>
>>>> >>>>>>> Cheers,
>>>> >>>>>>> Max
>>>> >>>>>>>
>>>> >>>>>>> On 21.05.19 16:49, Jan Lukavský wrote:
>>>> >>>>>>>> Hi,
>>>> >>>>>>>>
>>>> >>>>>>>> > Actually, I think it is a larger (open) question whether
>>>> exactly
>>>> >>>>>>>> once is guaranteed by the model or whether runners are allowed
>>>> to
>>>> >>>>>>>> relax that. I would think, however, that sources correctly
>>>> >>>>>>>> implemented should be idempotent when run atop an exactly once
>>>> >>>>>>>> infrastructure such as Flink of Dataflow.
>>>> >>>>>>>>
>>>> >>>>>>>> I would assume, that the model basically inherits guarantees of
>>>> >>>>>>>> underlying infrastructure. Because Flink does not work as you
>>>> >>>>>>>> described (atomic commit of inputs, state and outputs), but
>>>> rather a
>>>> >>>>>>>> checkpoint mark is flowing through the DAG much like watermark
>>>> and on
>>>> >>>>>>>> failures operators are restored and data reprocessed, it (IMHO)
>>>> >>>>>>>> implies, that you have exactly once everywhere in the DAG *but*
>>>> >>>>>>>> sinks. That is because sinks cannot be restored to previous
>>>> state,
>>>> >>>>>>>> instead sinks are supposed to be idempotent in order for the
>>>> exactly
>>>> >>>>>>>> once to really work (or at least be able to commit outputs on
>>>> >>>>>>>> checkpoint in sink). That implies that if you don't have sink
>>>> that is
>>>> >>>>>>>> able to commit outputs atomically on checkpoint, the pipeline
>>>> >>>>>>>> execution should be deterministic upon retries, otherwise
>>>> shadow
>>>> >>>>>>>> writes from failed paths of the pipeline might appear.
>>>> >>>>>>>>
>>>> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
>>>> >>>>>>>> current understanding.
>>>> >>>>>>>>
>>>> >>>>>>>> > Sounds like we should make this clearer.
>>>> >>>>>>>>
>>>> >>>>>>>> I meant that you are right that we must not in any thoughts we
>>>> are
>>>> >>>>>>>> having forget that streams are by definition out-of-order.
>>>> That is
>>>> >>>>>>>> property that we cannot change. But - that doesn't limit us
>>>> from
>>>> >>>>>>>> creating operator that presents the data to UDF as if the
>>>> stream was
>>>> >>>>>>>> ideally sorted. It can do that by introducing latency, of
>>>> course.
>>>> >>>>>>>>
>>>> >>>>>>>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>>>> >>>>>>>>> Reza: One could provide something like this as a utility
>>>> class, but
>>>> >>>>>>>>> one downside is that it is not scale invariant. It requires a
>>>> tuning
>>>> >>>>>>>>> parameter that, if to small, won't mitigate the problem, but
>>>> if to
>>>> >>>>>>>>> big, greatly increases latency. (Possibly one could define a
>>>> dynamic
>>>> >>>>>>>>> session-like window to solve this though...) It also might be
>>>> harder
>>>> >>>>>>>>> for runners that *can* cheaply present stuff in timestamp
>>>> order to
>>>> >>>>>>>>> optimize. (That and, in practice, our annotation-style
>>>> process methods
>>>> >>>>>>>>> don't lend themselves to easy composition.) I think it could
>>>> work in
>>>> >>>>>>>>> specific cases though.
>>>> >>>>>>>>>
>>>> >>>>>>>>> More inline below.
>>>> >>>>>>>>>
>>>> >>>>>>>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <
>>>> je.ik@seznam.cz> wrote:
>>>> >>>>>>>>>> Hi Robert,
>>>> >>>>>>>>>>
>>>> >>>>>>>>>> > Beam has an exactly-once model. If the data was
>>>> consumed, state
>>>> >>>>>>>>>> mutated, and outputs written downstream (these three are
>>>> committed
>>>> >>>>>>>>>> together atomically) it will not be replayed. That does not,
>>>> of
>>>> >>>>>>>>>> course,
>>>> >>>>>>>>>> solve the non-determanism due to ordering (including the
>>>> fact that two
>>>> >>>>>>>>>> operations reading the same PCollection may view different
>>>> ordering).
>>>> >>>>>>>>>>
>>>> >>>>>>>>>> I think what you describe is a property of a runner, not of
>>>> the model,
>>>> >>>>>>>>>> right? I think if I run my pipeline on Flink I will not get
>>>> this
>>>> >>>>>>>>>> atomicity, because although Flink uses also exactly-once
>>>> model if
>>>> >>>>>>>>>> might
>>>> >>>>>>>>>> write outputs multiple times.
>>>> >>>>>>>>> Actually, I think it is a larger (open) question whether
>>>> exactly once
>>>> >>>>>>>>> is guaranteed by the model or whether runners are allowed to
>>>> relax
>>>> >>>>>>>>> that. I would think, however, that sources correctly
>>>> implemented
>>>> >>>>>>>>> should be idempotent when run atop an exactly once
>>>> infrastructure such
>>>> >>>>>>>>> as Flink of Dataflow.
>>>> >>>>>>>>>
>>>> >>>>>>>>>> > 1) Is it correct for a (Stateful)DoFn to assume
>>>> elements are
>>>> >>>>>>>>>> received
>>>> >>>>>>>>>> in a specific order? In the current model, it is not. Being
>>>> able to
>>>> >>>>>>>>>> read, handle, and produced out-of-order data, including late
>>>> data,
>>>> >>>>>>>>>> is a
>>>> >>>>>>>>>> pretty fundamental property of distributed systems.
>>>> >>>>>>>>>>
>>>> >>>>>>>>>> Yes, absolutely. The argument here is not that Stateful
>>>> ParDo should
>>>> >>>>>>>>>> presume to receive elements in any order, but to _present_
>>>> it as
>>>> >>>>>>>>>> such to
>>>> >>>>>>>>>> the user @ProcessElement function.
>>>> >>>>>>>>> Sounds like we should make this clearer.
>>>> >>>>>>>>>
>>>> >>>>>>>>>> > 2) Given that some operations are easier (or possibly
>>>> only
>>>> >>>>>>>>>> possible)
>>>> >>>>>>>>>> to write when operating on ordered data, and that different
>>>> runners
>>>> >>>>>>>>>> may
>>>> >>>>>>>>>> have (significantly) cheaper ways to provide this ordering
>>>> than can be
>>>> >>>>>>>>>> done by the user themselves, should we elevate this to a
>>>> property of
>>>> >>>>>>>>>> (Stateful?)DoFns that the runner can provide? I think a
>>>> compelling
>>>> >>>>>>>>>> argument can be made here that we should.
>>>> >>>>>>>>>>
>>>> >>>>>>>>>> +1
>>>> >>>>>>>>>>
>>>> >>>>>>>>>> Jan
>>>> >>>>>>>>>>
>>>> >>>>>>>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>>>> >>>>>>>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <
>>>> je.ik@seznam.cz> wrote:
>>>> >>>>>>>>>>>> > I don't see batch vs. streaming as part of the
>>>> model. One
>>>> >>>>>>>>>>>> can have
>>>> >>>>>>>>>>>> microbatch, or even a runner that alternates between
>>>> different
>>>> >>>>>>>>>>>> modes.
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> Although I understand motivation of this statement, this
>>>> project
>>>> >>>>>>>>>>>> name is
>>>> >>>>>>>>>>>> "Apache Beam: An advanced unified programming model". What
>>>> does the
>>>> >>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the
>>>> model?
>>>> >>>>>>>>>>> What I mean is that streaming vs. batch is no longer part
>>>> of the
>>>> >>>>>>>>>>> model
>>>> >>>>>>>>>>> (or ideally API), but pushed down to be a concern of the
>>>> runner
>>>> >>>>>>>>>>> (executor) of the pipeline.
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <
>>>> je.ik@seznam.cz>
>>>> >>>>>>>>>>> wrote:
>>>> >>>>>>>>>>>> Hi Kenn,
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> OK, so if we introduce annotation, we can have stateful
>>>> ParDo
>>>> >>>>>>>>>>>> with sorting, that would perfectly resolve my issues. I
>>>> still
>>>> >>>>>>>>>>>> have some doubts, though. Let me explain. The current
>>>> behavior of
>>>> >>>>>>>>>>>> stateful ParDo has the following properties:
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> a) might fail in batch, although runs fine in
>>>> streaming (that
>>>> >>>>>>>>>>>> is due to the buffering, and unbounded lateness in batch,
>>>> which
>>>> >>>>>>>>>>>> was discussed back and forth in this thread)
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> b) might be non deterministic (this is because the
>>>> elements
>>>> >>>>>>>>>>>> arrive at somewhat random order, and even if you do the
>>>> operation
>>>> >>>>>>>>>>>> "assign unique ID to elements" this might produce different
>>>> >>>>>>>>>>>> results when run multiple times)
>>>> >>>>>>>>>>> PCollections are *explicitly* unordered. Any operations that
>>>> >>>>>>>>>>> assume or
>>>> >>>>>>>>>>> depend on a specific ordering for correctness (or
>>>> determinism) must
>>>> >>>>>>>>>>> provide that ordering themselves (i.e. tolerate "arbitrary
>>>> shuffling
>>>> >>>>>>>>>>> of inputs"). As you point out, that may be very expensive
>>>> if you have
>>>> >>>>>>>>>>> very hot keys with very large (unbounded) timestamp skew.
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>> StatefulDoFns are low-level operations that should be used
>>>> with care;
>>>> >>>>>>>>>>> the simpler windowing model gives determinism in the face of
>>>> >>>>>>>>>>> unordered
>>>> >>>>>>>>>>> data (though late data and non-end-of-window triggering
>>>> introduces
>>>> >>>>>>>>>>> some of the non-determanism back in).
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>>> What worries me most is the property b), because it seems
>>>> to me
>>>> >>>>>>>>>>>> to have serious consequences - not only that if you run
>>>> twice
>>>> >>>>>>>>>>>> batch pipeline you would get different results, but even on
>>>> >>>>>>>>>>>> streaming, when pipeline fails and gets restarted from
>>>> >>>>>>>>>>>> checkpoint, produced output might differ from the previous
>>>> run
>>>> >>>>>>>>>>>> and data from the first run might have already been
>>>> persisted
>>>> >>>>>>>>>>>> into sink. That would create somewhat messy outputs.
>>>> >>>>>>>>>>> Beam has an exactly-once model. If the data was consumed,
>>>> state
>>>> >>>>>>>>>>> mutated, and outputs written downstream (these three are
>>>> committed
>>>> >>>>>>>>>>> together atomically) it will not be replayed. That does
>>>> not, of
>>>> >>>>>>>>>>> course, solve the non-determanism due to ordering
>>>> (including the fact
>>>> >>>>>>>>>>> that two operations reading the same PCollection may view
>>>> different
>>>> >>>>>>>>>>> ordering).
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>>> These two properties makes me think that the current
>>>> >>>>>>>>>>>> implementation is more of a _special case_ than the
>>>> general one.
>>>> >>>>>>>>>>>> The general one would be that your state doesn't have the
>>>> >>>>>>>>>>>> properties to be able to tolerate buffering problems and/or
>>>> >>>>>>>>>>>> non-determinism. Which is the case where you need sorting
>>>> in both
>>>> >>>>>>>>>>>> streaming and batch to be part of the model.
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> Let me point out one more analogy - that is merging vs.
>>>> >>>>>>>>>>>> non-merging windows. The general case (merging windows)
>>>> implies
>>>> >>>>>>>>>>>> sorting by timestamp in both batch case (explicit) and
>>>> streaming
>>>> >>>>>>>>>>>> (buffering). The special case (non-merging windows)
>>>> doesn't rely
>>>> >>>>>>>>>>>> on any timestamp ordering, so the sorting and buffering
>>>> can be
>>>> >>>>>>>>>>>> dropped. The underlying root cause of this is the same for
>>>> both
>>>> >>>>>>>>>>>> stateful ParDo and windowing (essentially, assigning window
>>>> >>>>>>>>>>>> labels is a stateful operation when windowing function is
>>>> merging).
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> The reason for the current behavior of stateful ParDo
>>>> seems to be
>>>> >>>>>>>>>>>> performance, but is it right to abandon correctness in
>>>> favor of
>>>> >>>>>>>>>>>> performance? Wouldn't it be more consistent to have the
>>>> default
>>>> >>>>>>>>>>>> behavior prefer correctness and when you have the specific
>>>> >>>>>>>>>>>> conditions of state function having special properties,
>>>> then you
>>>> >>>>>>>>>>>> can annotate your DoFn (with something like
>>>> >>>>>>>>>>>> @TimeOrderingAgnostic), which would yield a better
>>>> performance in
>>>> >>>>>>>>>>>> that case?
>>>> >>>>>>>>>>> There are two separable questions here.
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>> 1) Is it correct for a (Stateful)DoFn to assume elements
>>>> are received
>>>> >>>>>>>>>>> in a specific order? In the current model, it is not. Being
>>>> able to
>>>> >>>>>>>>>>> read, handle, and produced out-of-order data, including
>>>> late data, is
>>>> >>>>>>>>>>> a pretty fundamental property of distributed systems.
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>> 2) Given that some operations are easier (or possibly only
>>>> possible)
>>>> >>>>>>>>>>> to write when operating on ordered data, and that different
>>>> runners
>>>> >>>>>>>>>>> may have (significantly) cheaper ways to provide this
>>>> ordering than
>>>> >>>>>>>>>>> can be done by the user themselves, should we elevate this
>>>> to a
>>>> >>>>>>>>>>> property of (Stateful?)DoFns that the runner can provide? I
>>>> think a
>>>> >>>>>>>>>>> compelling argument can be made here that we should.
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>> - Robert
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>>
>>>> >>>>>>>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> Thanks for the nice small example of a calculation that
>>>> depends
>>>> >>>>>>>>>>>> on order. You are right that many state machines have this
>>>> >>>>>>>>>>>> property. I agree w/ you and Luke that it is convenient
>>>> for batch
>>>> >>>>>>>>>>>> processing to sort by event timestamp before running a
>>>> stateful
>>>> >>>>>>>>>>>> ParDo. In streaming you could also implement "sort by event
>>>> >>>>>>>>>>>> timestamp" by buffering until you know all earlier data
>>>> will be
>>>> >>>>>>>>>>>> dropped - a slack buffer up to allowed lateness.
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> I do not think that it is OK to sort in batch and not in
>>>> >>>>>>>>>>>> streaming. Many state machines diverge very rapidly when
>>>> things
>>>> >>>>>>>>>>>> are out of order. So each runner if they see the
>>>> >>>>>>>>>>>> "@OrderByTimestamp" annotation (or whatever) needs to
>>>> deliver
>>>> >>>>>>>>>>>> sorted data (by some mix of buffering and dropping), or to
>>>> reject
>>>> >>>>>>>>>>>> the pipeline as unsupported.
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> And also want to say that this is not the default case -
>>>> many
>>>> >>>>>>>>>>>> uses of state & timers in ParDo yield different results at
>>>> the
>>>> >>>>>>>>>>>> element level, but the results are equivalent at in the big
>>>> >>>>>>>>>>>> picture. Such as the example of "assign a unique sequence
>>>> number
>>>> >>>>>>>>>>>> to each element" or "group into batches" it doesn't matter
>>>> >>>>>>>>>>>> exactly what the result is, only that it meets the spec.
>>>> And
>>>> >>>>>>>>>>>> other cases like user funnels are monotonic enough that
>>>> you also
>>>> >>>>>>>>>>>> don't actually need sorting.
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> Kenn
>>>> >>>>>>>>>>>>
>>>> >>>>>>>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <
>>>> je.ik@seznam.cz>
>>>> >>>>>>>>>>>> wrote:
>>>> >>>>>>>>>>>>> Yes, the problem will arise probably mostly when you have
>>>> not
>>>> >>>>>>>>>>>>> well distributed keys (or too few keys). I'm really not
>>>> sure if
>>>> >>>>>>>>>>>>> a pure GBK with a trigger can solve this - it might help
>>>> to have
>>>> >>>>>>>>>>>>> data driven trigger. There would still be some doubts,
>>>> though.
>>>> >>>>>>>>>>>>> The main question is still here - people say, that
>>>> sorting by
>>>> >>>>>>>>>>>>> timestamp before stateful ParDo would be prohibitively
>>>> slow, but
>>>> >>>>>>>>>>>>> I don't really see why - the sorting is very probably
>>>> already
>>>> >>>>>>>>>>>>> there. And if not (hash grouping instead of sorted
>>>> grouping),
>>>> >>>>>>>>>>>>> then the sorting would affect only user defined
>>>> StatefulParDos.
>>>> >>>>>>>>>>>>>
>>>> >>>>>>>>>>>>> This would suggest that the best way out of this would be
>>>> really
>>>> >>>>>>>>>>>>> to add annotation, so that the author of the pipeline can
>>>> decide.
>>>> >>>>>>>>>>>>>
>>>> >>>>>>>>>>>>> If that would be acceptable I think I can try to prepare
>>>> some
>>>> >>>>>>>>>>>>> basic functionality, but I'm not sure, if I would be able
>>>> to
>>>> >>>>>>>>>>>>> cover all runners / sdks.
>>>> >>>>>>>>>>>>>
>>>> >>>>>>>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>>>> >>>>>>>>>>>>>
>>>> >>>>>>>>>>>>> It is read all per key and window and not just read all
>>>> (this
>>>> >>>>>>>>>>>>> still won't scale with hot keys in the global window).
>>>> The GBK
>>>> >>>>>>>>>>>>> preceding the StatefulParDo will guarantee that you are
>>>> >>>>>>>>>>>>> processing all the values for a specific key and window
>>>> at any
>>>> >>>>>>>>>>>>> given time. Is there a specific window/trigger that is
>>>> missing
>>>> >>>>>>>>>>>>> that you feel would remove the need for you to use
>>>> StatefulParDo?
>>>> >>>>>>>>>>>>>
>>>> >>>>>>>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <
>>>> je.ik@seznam.cz>
>>>> >>>>>>>>>>>>> wrote:
>>>> >>>>>>>>>>>>>> Hi Lukasz,
>>>> >>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> Today, if you must have a strict order, you must
>>>> guarantee
>>>> >>>>>>>>>>>>>>> that your StatefulParDo implements the necessary
>>>> "buffering &
>>>> >>>>>>>>>>>>>>> sorting" into state.
>>>> >>>>>>>>>>>>>> Yes, no problem with that. But this whole discussion
>>>> started,
>>>> >>>>>>>>>>>>>> because *this doesn't work on batch*. You simply cannot
>>>> first
>>>> >>>>>>>>>>>>>> read everything from distributed storage and then buffer
>>>> it all
>>>> >>>>>>>>>>>>>> into memory, just to read it again, but sorted. That
>>>> will not
>>>> >>>>>>>>>>>>>> work. And even if it would, it would be a terrible waste
>>>> of
>>>> >>>>>>>>>>>>>> resources.
>>>> >>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>> Jan
>>>> >>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>>>> >>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <
>>>> je.ik@seznam.cz>
>>>> >>>>>>>>>>>>>> wrote:
>>>> >>>>>>>>>>>>>>> This discussion brings many really interesting
>>>> questions for
>>>> >>>>>>>>>>>>>>> me. :-)
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> > I don't see batch vs. streaming as part of the
>>>> model. One
>>>> >>>>>>>>>>>>>>> can have
>>>> >>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
>>>> different
>>>> >>>>>>>>>>>>>>> modes.
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> Although I understand motivation of this statement, this
>>>> >>>>>>>>>>>>>>> project name is
>>>> >>>>>>>>>>>>>>> "Apache Beam: An advanced unified programming model".
>>>> What
>>>> >>>>>>>>>>>>>>> does the
>>>> >>>>>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of
>>>> the model?
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> Using microbatching, chaining of batch jobs, or pure
>>>> streaming
>>>> >>>>>>>>>>>>>>> are
>>>> >>>>>>>>>>>>>>> exactly the "runtime conditions/characteristics" I
>>>> refer to.
>>>> >>>>>>>>>>>>>>> All these
>>>> >>>>>>>>>>>>>>> define several runtime parameters, which in turn define
>>>> how
>>>> >>>>>>>>>>>>>>> well/badly
>>>> >>>>>>>>>>>>>>> will the pipeline perform and how many resources might
>>>> be
>>>> >>>>>>>>>>>>>>> needed. From
>>>> >>>>>>>>>>>>>>> my point of view, pure streaming should be the most
>>>> resource
>>>> >>>>>>>>>>>>>>> demanding
>>>> >>>>>>>>>>>>>>> (if not, why bother with batch? why not run everything
>>>> in
>>>> >>>>>>>>>>>>>>> streaming
>>>> >>>>>>>>>>>>>>> only? what will there remain to "unify"?).
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> > Fortunately, for batch, only the state for a
>>>> single key
>>>> >>>>>>>>>>>>>>> needs to be
>>>> >>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys
>>>> across
>>>> >>>>>>>>>>>>>>> the range
>>>> >>>>>>>>>>>>>>> of skew. Of course if you have few or hot keys, one can
>>>> still
>>>> >>>>>>>>>>>>>>> have
>>>> >>>>>>>>>>>>>>> issues (and this is not specific to StatefulDoFns).
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> Yes, but here is still the presumption that my stateful
>>>> DoFn can
>>>> >>>>>>>>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain
>>>> the use
>>>> >>>>>>>>>>>>>>> case in
>>>> >>>>>>>>>>>>>>> more detail.
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> Suppose you have input stream consisting of 1s and 0s
>>>> (and
>>>> >>>>>>>>>>>>>>> some key for
>>>> >>>>>>>>>>>>>>> each element, which is irrelevant for the
>>>> demonstration). Your
>>>> >>>>>>>>>>>>>>> task is
>>>> >>>>>>>>>>>>>>> to calculate in running global window the actual number
>>>> of
>>>> >>>>>>>>>>>>>>> changes
>>>> >>>>>>>>>>>>>>> between state 0 and state 1 and vice versa. When the
>>>> state
>>>> >>>>>>>>>>>>>>> doesn't
>>>> >>>>>>>>>>>>>>> change, you don't calculate anything. If input (for
>>>> given key)
>>>> >>>>>>>>>>>>>>> would be
>>>> >>>>>>>>>>>>>>> (tN denotes timestamp N):
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t1: 1
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t2: 0
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t3: 0
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t4: 1
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t5: 1
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t6: 0
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> then the output should yield (supposing that default
>>>> state is
>>>> >>>>>>>>>>>>>>> zero):
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t1: (one: 1, zero: 0)
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t2: (one: 1, zero: 1)
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t3: (one: 1, zero: 1)
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t4: (one: 2, zero: 1)
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t5: (one: 2, zero: 1)
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> t6: (one: 2, zero: 2)
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> How would you implement this in current Beam semantics?
>>>> >>>>>>>>>>>>>> I think your saying here that I know that my input is
>>>> ordered
>>>> >>>>>>>>>>>>>> in a specific way and since I assume the order when
>>>> writing my
>>>> >>>>>>>>>>>>>> pipeline I can perform this optimization. But there is
>>>> nothing
>>>> >>>>>>>>>>>>>> preventing a runner from noticing that your processing
>>>> in the
>>>> >>>>>>>>>>>>>> global window with a specific type of trigger and
>>>> re-ordering
>>>> >>>>>>>>>>>>>> your inputs/processing to get better performance (since
>>>> you
>>>> >>>>>>>>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>>>> >>>>>>>>>>>>>> streaming for the GlobalWindow).
>>>> >>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>> Today, if you must have a strict order, you must
>>>> guarantee that
>>>> >>>>>>>>>>>>>> your StatefulParDo implements the necessary "buffering &
>>>> >>>>>>>>>>>>>> sorting" into state. I can see why you would want an
>>>> annotation
>>>> >>>>>>>>>>>>>> that says I must have timestamp ordered elements, since
>>>> it
>>>> >>>>>>>>>>>>>> makes writing certain StatefulParDos much easier.
>>>> StatefulParDo
>>>> >>>>>>>>>>>>>> is a low-level function, it really is the "here you go
>>>> and do
>>>> >>>>>>>>>>>>>> whatever you need to but here be dragons" function while
>>>> >>>>>>>>>>>>>> windowing and triggering is meant to keep many people
>>>> from
>>>> >>>>>>>>>>>>>> writing StatefulParDo in the first place.
>>>> >>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> > Pipelines that fail in the "worst case" batch
>>>> scenario
>>>> >>>>>>>>>>>>>>> are likely to
>>>> >>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the
>>>> watermark
>>>> >>>>>>>>>>>>>>> falls
>>>> >>>>>>>>>>>>>>> behind in streaming mode as well.
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> But the worst case is defined by input of size
>>>> (available
>>>> >>>>>>>>>>>>>>> resources +
>>>> >>>>>>>>>>>>>>> single byte) -> pipeline fail. Although it could have
>>>> >>>>>>>>>>>>>>> finished, given
>>>> >>>>>>>>>>>>>>> the right conditions.
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> > This might be reasonable, implemented by default
>>>> by
>>>> >>>>>>>>>>>>>>> buffering
>>>> >>>>>>>>>>>>>>> everything and releasing elements as the watermark
>>>> (+lateness)
>>>> >>>>>>>>>>>>>>> advances,
>>>> >>>>>>>>>>>>>>> but would likely lead to inefficient (though *maybe*
>>>> easier to
>>>> >>>>>>>>>>>>>>> reason
>>>> >>>>>>>>>>>>>>> about) code.
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> Sure, the pipeline will be less efficient, because it
>>>> would
>>>> >>>>>>>>>>>>>>> have to
>>>> >>>>>>>>>>>>>>> buffer and sort the inputs. But at least it will produce
>>>> >>>>>>>>>>>>>>> correct results
>>>> >>>>>>>>>>>>>>> in cases where updates to state are order-sensitive.
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> > Would it be roughly equivalent to GBK +
>>>> FlatMap(lambda
>>>> >>>>>>>>>>>>>>> (key, values):
>>>> >>>>>>>>>>>>>>> [(key, value) for value in values])?
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> I'd say roughly yes, but difference would be in the
>>>> trigger.
>>>> >>>>>>>>>>>>>>> The trigger
>>>> >>>>>>>>>>>>>>> should ideally fire as soon as watermark (+lateness)
>>>> crosses
>>>> >>>>>>>>>>>>>>> element
>>>> >>>>>>>>>>>>>>> with lowest timestamp in the buffer. Although this
>>>> could be
>>>> >>>>>>>>>>>>>>> somehow
>>>> >>>>>>>>>>>>>>> emulated by fixed trigger each X millis.
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> > Or is the underlying desire just to be able to
>>>> hint to
>>>> >>>>>>>>>>>>>>> the runner
>>>> >>>>>>>>>>>>>>> that the code may perform better (e.g. require less
>>>> resources)
>>>> >>>>>>>>>>>>>>> as skew
>>>> >>>>>>>>>>>>>>> is reduced (and hence to order by timestamp iff it's
>>>> cheap)?
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> No, the sorting would have to be done in streaming case
>>>> as
>>>> >>>>>>>>>>>>>>> well. That is
>>>> >>>>>>>>>>>>>>> an imperative of the unified model. I think it is
>>>> possible to
>>>> >>>>>>>>>>>>>>> sort by
>>>> >>>>>>>>>>>>>>> timestamp only in batch case (and do it for *all* batch
>>>> >>>>>>>>>>>>>>> stateful pardos
>>>> >>>>>>>>>>>>>>> without annotation), or introduce annotation, but then
>>>> make
>>>> >>>>>>>>>>>>>>> the same
>>>> >>>>>>>>>>>>>>> guarantees for streaming case as well.
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> Jan
>>>> >>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>>>> >>>>>>>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>>>> >>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>>>> >>>>>>>>>>>>>>>>> Hi Robert,
>>>> >>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>> yes, I think you rephrased my point - although no
>>>> *explicit*
>>>> >>>>>>>>>>>>>>>>> guarantees
>>>> >>>>>>>>>>>>>>>>> of ordering are given in either mode, there is
>>>> *implicit*
>>>> >>>>>>>>>>>>>>>>> ordering in
>>>> >>>>>>>>>>>>>>>>> streaming case that is due to nature of the
>>>> processing - the
>>>> >>>>>>>>>>>>>>>>> difference
>>>> >>>>>>>>>>>>>>>>> between watermark and timestamp of elements flowing
>>>> through
>>>> >>>>>>>>>>>>>>>>> the pipeline
>>>> >>>>>>>>>>>>>>>>> are generally low (too high difference leads to the
>>>> >>>>>>>>>>>>>>>>> overbuffering
>>>> >>>>>>>>>>>>>>>>> problem), but there is no such bound on batch.
>>>> >>>>>>>>>>>>>>>> Fortunately, for batch, only the state for a single
>>>> key needs
>>>> >>>>>>>>>>>>>>>> to be
>>>> >>>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys
>>>> >>>>>>>>>>>>>>>> across the
>>>> >>>>>>>>>>>>>>>> range of skew. Of course if you have few or hot keys,
>>>> one can
>>>> >>>>>>>>>>>>>>>> still
>>>> >>>>>>>>>>>>>>>> have issues (and this is not specific to
>>>> StatefulDoFns).
>>>> >>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>> As a result, I see a few possible solutions:
>>>> >>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>> - the best and most natural seems to be
>>>> extension of
>>>> >>>>>>>>>>>>>>>>> the model, so
>>>> >>>>>>>>>>>>>>>>> that it defines batch as not only "streaming pipeline
>>>> >>>>>>>>>>>>>>>>> executed in batch
>>>> >>>>>>>>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>>>> >>>>>>>>>>>>>>>>> characteristics as
>>>> >>>>>>>>>>>>>>>>> in streaming case, executed in batch fashion", I
>>>> really
>>>> >>>>>>>>>>>>>>>>> don't think that
>>>> >>>>>>>>>>>>>>>>> there are any conflicts with the current model, or
>>>> that this
>>>> >>>>>>>>>>>>>>>>> could
>>>> >>>>>>>>>>>>>>>>> affect performance, because the required sorting (as
>>>> pointed by
>>>> >>>>>>>>>>>>>>>>> Aljoscha) is very probably already done during
>>>> translation
>>>> >>>>>>>>>>>>>>>>> of stateful
>>>> >>>>>>>>>>>>>>>>> pardos. Also note that this definition only affects
>>>> user
>>>> >>>>>>>>>>>>>>>>> defined
>>>> >>>>>>>>>>>>>>>>> stateful pardos
>>>> >>>>>>>>>>>>>>>> I don't see batch vs. streaming as part of the model.
>>>> One can
>>>> >>>>>>>>>>>>>>>> have
>>>> >>>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
>>>> >>>>>>>>>>>>>>>> different modes.
>>>> >>>>>>>>>>>>>>>> The model describes what the valid outputs are given a
>>>> >>>>>>>>>>>>>>>> (sometimes
>>>> >>>>>>>>>>>>>>>> partial) set of inputs. It becomes really hard to
>>>> define
>>>> >>>>>>>>>>>>>>>> things like
>>>> >>>>>>>>>>>>>>>> "as good runtime characteristics." Once you allow any
>>>> >>>>>>>>>>>>>>>> out-of-orderedness, it is not very feasible to try and
>>>> define
>>>> >>>>>>>>>>>>>>>> (and
>>>> >>>>>>>>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>>>> >>>>>>>>>>>>>>>> out-of-orderedness.
>>>> >>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>> Pipelines that fail in the "worst case" batch scenario
>>>> are
>>>> >>>>>>>>>>>>>>>> likely to
>>>> >>>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the
>>>> watermark
>>>> >>>>>>>>>>>>>>>> falls
>>>> >>>>>>>>>>>>>>>> behind in streaming mode as well.
>>>> >>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>> - another option would be to introduce
>>>> annotation for
>>>> >>>>>>>>>>>>>>>>> DoFns (e.g.
>>>> >>>>>>>>>>>>>>>>> @RequiresStableTimeCharacteristics), which would
>>>> result in
>>>> >>>>>>>>>>>>>>>>> the sorting
>>>> >>>>>>>>>>>>>>>>> in batch case - but - this extension would have to
>>>> ensure
>>>> >>>>>>>>>>>>>>>>> the sorting in
>>>> >>>>>>>>>>>>>>>>> streaming mode also - it would require definition of
>>>> allowed
>>>> >>>>>>>>>>>>>>>>> lateness,
>>>> >>>>>>>>>>>>>>>>> and triggger (essentially similar to window)
>>>> >>>>>>>>>>>>>>>> This might be reasonable, implemented by default by
>>>> buffering
>>>> >>>>>>>>>>>>>>>> everything and releasing elements as the watermark
>>>> (+lateness)
>>>> >>>>>>>>>>>>>>>> advances, but would likely lead to inefficient (though
>>>> >>>>>>>>>>>>>>>> *maybe* easier
>>>> >>>>>>>>>>>>>>>> to reason about) code. Not sure about the semantics of
>>>> >>>>>>>>>>>>>>>> triggering
>>>> >>>>>>>>>>>>>>>> here, especially data-driven triggers. Would it be
>>>> roughly
>>>> >>>>>>>>>>>>>>>> equivalent
>>>> >>>>>>>>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value)
>>>> for
>>>> >>>>>>>>>>>>>>>> value in
>>>> >>>>>>>>>>>>>>>> values])?
>>>> >>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>> Or is the underlying desire just to be able to hint to
>>>> the
>>>> >>>>>>>>>>>>>>>> runner that
>>>> >>>>>>>>>>>>>>>> the code may perform better (e.g. require less
>>>> resources) as
>>>> >>>>>>>>>>>>>>>> skew is
>>>> >>>>>>>>>>>>>>>> reduced (and hence to order by timestamp iff it's
>>>> cheap)?
>>>> >>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>> - last option would be to introduce these
>>>> "higher order
>>>> >>>>>>>>>>>>>>>>> guarantees" in
>>>> >>>>>>>>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to
>>>> be the
>>>> >>>>>>>>>>>>>>>>> worst
>>>> >>>>>>>>>>>>>>>>> option to me
>>>> >>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>> I see the first two options quite equally good,
>>>> although the
>>>> >>>>>>>>>>>>>>>>> letter one
>>>> >>>>>>>>>>>>>>>>> is probably more time consuming to implement. But it
>>>> would
>>>> >>>>>>>>>>>>>>>>> bring
>>>> >>>>>>>>>>>>>>>>> additional feature to streaming case as well.
>>>> >>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>> Thanks for any thoughts.
>>>> >>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>> Jan
>>>> >>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>>>> >>>>>>>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>>>> >>>>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>>>> >>>>>>>>>>>>>>>>>>> Hi Reuven,
>>>> >>>>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in
>>>> batch
>>>> >>>>>>>>>>>>>>>>>>>> runners.
>>>> >>>>>>>>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic
>>>> inside
>>>> >>>>>>>>>>>>>>>>>>> the state works for absolutely unbounded
>>>> out-of-orderness
>>>> >>>>>>>>>>>>>>>>>>> of elements. That basically (practically) can work
>>>> only
>>>> >>>>>>>>>>>>>>>>>>> for cases, where the order of input elements doesn't
>>>> >>>>>>>>>>>>>>>>>>> matter. But, "state" can refer to "state machine",
>>>> and any
>>>> >>>>>>>>>>>>>>>>>>> time you have a state machine involved, then the
>>>> ordering
>>>> >>>>>>>>>>>>>>>>>>> of elements would matter.
>>>> >>>>>>>>>>>>>>>>>> No guarantees on order are provided in *either*
>>>> streaming
>>>> >>>>>>>>>>>>>>>>>> or batch
>>>> >>>>>>>>>>>>>>>>>> mode by the model. However, it is the case that in
>>>> order to
>>>> >>>>>>>>>>>>>>>>>> make
>>>> >>>>>>>>>>>>>>>>>> forward progress most streaming runners attempt to
>>>> limit
>>>> >>>>>>>>>>>>>>>>>> the amount of
>>>> >>>>>>>>>>>>>>>>>> out-of-orderedness of elements (in terms of event
>>>> time vs.
>>>> >>>>>>>>>>>>>>>>>> processing
>>>> >>>>>>>>>>>>>>>>>> time) to make forward progress, which in turn could
>>>> help
>>>> >>>>>>>>>>>>>>>>>> cap the
>>>> >>>>>>>>>>>>>>>>>> amount of state that must be held concurrently,
>>>> whereas a
>>>> >>>>>>>>>>>>>>>>>> batch runner
>>>> >>>>>>>>>>>>>>>>>> may not allow any state to be safely discarded until
>>>> the whole
>>>> >>>>>>>>>>>>>>>>>> timeline from infinite past to infinite future has
>>>> been
>>>> >>>>>>>>>>>>>>>>>> observed.
>>>> >>>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>>> Also, as pointed out, state is not preserved "batch
>>>> to
>>>> >>>>>>>>>>>>>>>>>> batch" in batch mode.
>>>> >>>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>>>> >>>>>>>>>>>>>>>>>> <mx...@apache.org> wrote:
>>>> >>>>>>>>>>>>>>>>>>
>>>> >>>>>>>>>>>>>>>>>>>> batch semantics and streaming semantics
>>>> differs only
>>>> >>>>>>>>>>>>>>>>>>>> in that I can have GlobalWindow with default
>>>> trigger on
>>>> >>>>>>>>>>>>>>>>>>>> batch and cannot on stream
>>>> >>>>>>>>>>>>>>>>>>> You can have a GlobalWindow in streaming with a
>>>> default
>>>> >>>>>>>>>>>>>>>>>>> trigger. You
>>>> >>>>>>>>>>>>>>>>>>> could define additional triggers that do early
>>>> firings.
>>>> >>>>>>>>>>>>>>>>>>> And you could
>>>> >>>>>>>>>>>>>>>>>>> even trigger the global window by advancing the
>>>> watermark
>>>> >>>>>>>>>>>>>>>>>>> to +inf.
>>>> >>>>>>>>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global
>>>> window with
>>>> >>>>>>>>>>>>>>>>>> default
>>>> >>>>>>>>>>>>>>>>>> trigger on unbounded PCollections in the SDK because
>>>> this
>>>> >>>>>>>>>>>>>>>>>> is more
>>>> >>>>>>>>>>>>>>>>>> likely to be user error than an actual desire to
>>>> have no
>>>> >>>>>>>>>>>>>>>>>> output until
>>>> >>>>>>>>>>>>>>>>>> drain. But it's semantically valid in the model.
>>>>
>>>>
Re: Definition of Unified model
Posted by Jan Lukavský <je...@seznam.cz>.
Hi Lukasz, do you suggest, that ordering can have some relation to
iterations? That sounds interesting, and from the top of my head, it can
be true - you have to order on iteration number to get consistent
results. Maybe we are tracking something deeper here! :-)
Jan
On 5/29/19 7:02 PM, Lukasz Cwik wrote:
> Expanding the dimensionality could be the basis for loops within the
> graph since loops could be modeled as (time, loop iteration #, nested
> loop iteration #, nested nested loop iteration #, ...)
>
> On Tue, May 28, 2019 at 12:10 PM Jan Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
>
> Could this be solved by "expanding the dimensionality" of time
> field? What I mean by that - if input element to to FlatMap has
> sequence number T, then the (stateless) FlatMap knows the ordering
> of output elements, right? If it would expand the field by which
> it will next sort the elements to (X, 1), (X, 2), ... (X, N), then
> it would be possible to sort the elements back later. There seems
> to be no need for state to achieve that, or?
>
> Jan
>
> On 5/28/19 6:52 PM, Reuven Lax wrote:
>> A slightly larger concern: it also will force users to create
>> stateful DoFns everywhere to generate these sequence numbers. If
>> I have a ParDo that is not a simple 1:1 transform (i.e. not
>> MapElements), then the ParDo will need to generate its own
>> sequence numbers for ordering, and the only safe way to do so is
>> to use a stateful DoFn. This turns what used to be a simple
>> in-memory DoFn into one that has to access state. Also I believe
>> many runners will not fuse stateful DoFns. While none of this
>> poses a problem for the model, it could make ordering extremely
>> expensive to achieve.
>>
>> Reuven
>>
>> On Tue, May 28, 2019 at 6:09 AM Jan Lukavský <je.ik@seznam.cz
>> <ma...@seznam.cz>> wrote:
>>
>> Hi Reuven,
>>
>> > It also gets awkward with Flatten - the sequence number is
>> no longer enough, you must also encode which side of the
>> flatten each element came from.
>>
>> That is a generic need. Even if you read data from Kafka, the
>> offsets are comparable only inside single partition. So, for
>> Kafka to work as a FIFO for ordering, elements with same key
>> have to be pushed to the same partition (otherwise Kafka
>> cannot act as FIFO, because different partitions can be
>> handled by different brokers, which means different observers
>> and they therefore might not agree on the order of events).
>> So if we want to emulate FIFO per key, then the sequence IDs
>> have also be per key.
>>
>> On 5/28/19 2:33 PM, Reuven Lax wrote:
>>> Sequence metadata does have the disadvantage that users can
>>> no longer use the types coming from the source. You must
>>> create a new type that contains a sequence number (unless
>>> Beam provides this). It also gets awkward with Flatten - the
>>> sequence number is no longer enough, you must also encode
>>> which side of the flatten each element came from.
>>>
>>> On Tue, May 28, 2019 at 3:18 AM Jan Lukavský
>>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>>>
>>> As I understood it, Kenn was supporting the idea that
>>> sequence metadata
>>> is preferable over FIFO. I was trying to point out, that
>>> it even should
>>> provide the same functionally as FIFO, plus one
>>> important more -
>>> reproducibility and ability to being persisted and
>>> reused the same way
>>> in batch and streaming.
>>>
>>> There is no doubt, that sequence metadata can be stored
>>> in every
>>> storage. But, regarding some implicit ordering that
>>> sources might have -
>>> yes, of course, data written into HDFS or Cloud Storage
>>> has ordering,
>>> but only partial - inside some bulk (e.g. file) and the
>>> ordering is not
>>> defined correctly on boundaries of these bulks (between
>>> files). That is
>>> why I'd say, that ordering of sources is relevant only for
>>> (partitioned!) streaming sources and generally always
>>> reduces to
>>> sequence metadata (e.g. offsets).
>>>
>>> Jan
>>>
>>> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
>>> > Huge +1 to all Kenn said.
>>> >
>>> > Jan, batch sources can have orderings too, just like
>>> Kafka. I think
>>> > it's reasonable (for both batch and streaming) that if
>>> a source has an
>>> > ordering that is an important part of the data, it
>>> should preserve
>>> > this ordering into the data itself (e.g. as sequence
>>> numbers, offsets,
>>> > etc.)
>>> >
>>> > On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles
>>> <kenn@apache.org <ma...@apache.org>> wrote:
>>> >> I strongly prefer explicit sequence metadata over
>>> FIFO requirements, because:
>>> >>
>>> >> - FIFO is complex to specify: for example Dataflow
>>> has "per stage key-to-key" FIFO today, but it is not
>>> guaranteed to remain so (plus "stage" is not a portable
>>> concept, nor even guaranteed to remain a Dataflow concept)
>>> >> - complex specifications are by definition poor
>>> usability (if necessary, then it is what it is)
>>> >> - overly restricts the runner, reduces parallelism,
>>> for example any non-stateful ParDo has per-element
>>> parallelism, not per "key"
>>> >> - another perspective on that: FIFO makes everyone
>>> pay rather than just the transform that requires exactly
>>> sequencing
>>> >> - previous implementation details like reshuffles
>>> become part of the model
>>> >> - I'm not even convinced the use cases involved are
>>> addressed by some careful FIFO restrictions; many sinks
>>> re-key and they would all have to become aware of how
>>> keying of a sequence of "stages" affects the end-to-end FIFO
>>> >>
>>> >> A noop becoming a non-noop is essentially the
>>> mathematical definition of moving from higher-level to
>>> lower-level abstraction.
>>> >>
>>> >> So this strikes at the core question of what level of
>>> abstraction Beam aims to represent. Lower-level means
>>> there are fewer possible implementations and it is more
>>> tied to the underlying architecture, and anything not
>>> near-exact match pays a huge penalty. Higher-level means
>>> there are more implementations possible with different
>>> tradeoffs, though they may all pay a minor penalty.
>>> >>
>>> >> I could be convinced to change my mind, but it needs
>>> some extensive design, examples, etc. I think it is
>>> probably about the most consequential design decision in
>>> the whole Beam model, around the same level as the
>>> decision to use ParDo and GBK as the primitives IMO.
>>> >>
>>> >> Kenn
>>> >>
>>> >> On Thu, May 23, 2019 at 10:17 AM Reuven Lax
>>> <relax@google.com <ma...@google.com>> wrote:
>>> >>> Not really. I'm suggesting that some variant of FIFO
>>> ordering is necessary, which requires either runners
>>> natively support FIFO ordering or transforms adding some
>>> extra sequence number to each record to sort by.
>>> >>>
>>> >>> I still think your proposal is very useful by the
>>> way. I'm merely pointing out that to solve the
>>> state-machine problem we probably need something more.
>>> >>>
>>> >>> Reuven
>>> >>>
>>> >>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský
>>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>>> >>>> Hi,
>>> >>>> yes. It seems that ordering by user supplied UDF
>>> makes sense and I will update the design proposal
>>> accordingly.
>>> >>>> Would that solve the issues you mention?
>>> >>>> Jan
>>> >>>> ---------- Původní e-mail ----------
>>> >>>> Od: Reuven Lax <relax@google.com
>>> <ma...@google.com>>
>>> >>>> Komu: dev <dev@beam.apache.org
>>> <ma...@beam.apache.org>>
>>> >>>> Datum: 23. 5. 2019 18:44:38
>>> >>>> Předmět: Re: Definition of Unified model
>>> >>>>
>>> >>>> I'm simply saying that timestamp ordering is
>>> insufficient for state machines. I wasn't proposing
>>> Kafka as a solution - that was simply an example of how
>>> people solve this problem in other scenarios.
>>> >>>>
>>> >>>> BTW another example of ordering: Imagine today that
>>> you have a triggered Sum aggregation writing out to a
>>> key-value sink. In theory we provide no ordering, so the
>>> sink might write the triggered sums in the wrong order,
>>> ending up with an incorrect value in the sink. In this
>>> case you probably want values ordered by trigger pane index.
>>> >>>>
>>> >>>> Reuven
>>> >>>>
>>> >>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský
>>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>>> >>>>
>>> >>>> Hi Reuven,
>>> >>>> I share the view point of Robert. I think the isuue
>>> you refer to is not in reality related to timestamps,
>>> but to the fact, that ordering of events in time is
>>> observer dependent (either caused by relativity, or time
>>> skew, essentially this has the same consequences). And
>>> the resolution in fact isn't Kafka, but generally an
>>> authoritative observer, that tells you "I saw the events
>>> in this order". And you either have one (and have the
>>> outcome of his observation persisted in the data - e.g.
>>> as offset in Kafka partition), then you should be able
>>> to use it (maybe that suggests afterall that sorting by
>>> some user supplied UDF might make sense), or do not have
>>> it, and then any interpretation of the data seems to be
>>> equally valid. Although determinism is fine, of course.
>>> >>>> Jan
>>> >>>> ---------- Původní e-mail ----------
>>> >>>> Od: Reuven Lax <relax@google.com
>>> <ma...@google.com>>
>>> >>>> Komu: dev <dev@beam.apache.org
>>> <ma...@beam.apache.org>>
>>> >>>> Datum: 23. 5. 2019 17:39:12
>>> >>>> Předmět: Re: Definition of Unified model
>>> >>>>
>>> >>>> So an example would be elements of type
>>> "startUserSession" and "endUserSession" (website
>>> sessions, not Beam sessions). Logically you may need to
>>> process them in the correct order if you have any sort
>>> of state-machine logic. However timestamp ordering is
>>> never guaranteed to match the logical ordering. Not only
>>> might you have several elements with the same timestamp,
>>> but in reality time skew across backend servers can
>>> cause the events to have timestamps in reverse order of
>>> the actual causality order.
>>> >>>>
>>> >>>> People do solve this problem today though. Publish
>>> the events to Kafka, making sure that events for the
>>> same user end up in the same Kafka partition. This
>>> ensures that the events appear in the Kafka partitions
>>> in causality order, even if the timestamp order doesn't
>>> match. The your Kafka subscriber simply process the
>>> elements in each partition in order.
>>> >>>>
>>> >>>> I think the ability to impose FIFO causality
>>> ordering is what's needed for any state-machine work.
>>> Timestamp ordering has advantages (though often I think
>>> the advantage is in state), but does not solve this problem.
>>> >>>>
>>> >>>> Reuven
>>> >>>>
>>> >>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw
>>> <robertwb@google.com <ma...@google.com>> wrote:
>>> >>>>
>>> >>>> Good point.
>>> >>>>
>>> >>>> The "implementation-specific" way I would do this is
>>> >>>> window-by-instant, followed by a DoFn that gets all
>>> the elements with
>>> >>>> the same timestamp and sorts/acts accordingly, but
>>> this counts on the
>>> >>>> runner producing windows in timestamp order
>>> (likely?) and also the
>>> >>>> subsequent DoFn getting them in this order (also
>>> likely, due to
>>> >>>> fusion).
>>> >>>>
>>> >>>> One could make the argument that, though it does
>>> not provide
>>> >>>> deterministic behavior, getting elements of the
>>> same timestamp in
>>> >>>> different orders should produce equally valid
>>> interpretations of the
>>> >>>> data. (After all, due to relatively, timestamps are
>>> not technically
>>> >>>> well ordered across space.) I can see how
>>> data-dependent tiebreakers
>>> >>>> could be useful, or promises of preservation of
>>> order between
>>> >>>> operations.
>>> >>>>
>>> >>>> - Robert
>>> >>>>
>>> >>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax
>>> <relax@google.com <ma...@google.com>> wrote:
>>> >>>>> So Jan's example of state machines is quite a
>>> valid use case for ordering. However in my experience,
>>> timestamp ordering is insufficient for state machines.
>>> Elements that cause state transitions might come in with
>>> the exact same timestamp, yet still have a necessary
>>> ordering. Especially given Beam's decision to have
>>> milliseconds timestamps this is possible, but even at
>>> microsecond or nanosecond precision this can happen at
>>> scale. To handle state machines you usually need some
>>> sort of FIFO ordering along with an ordered sources,
>>> such as Kafka, not timestamp ordering.
>>> >>>>>
>>> >>>>> Reuven
>>> >>>>>
>>> >>>>> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský
>>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>>> >>>>>> Hi all,
>>> >>>>>>
>>> >>>>>> thanks everyone for this discussion. I think I
>>> have gathered enough
>>> >>>>>> feedback to be able to put down a proposition for
>>> changes, which I will
>>> >>>>>> do and send to this list for further discussion.
>>> There are still doubts
>>> >>>>>> remaining the non-determinism and it's relation
>>> to outputs stability vs.
>>> >>>>>> latency. But I will try to clarify all this in
>>> the design document.
>>> >>>>>>
>>> >>>>>> Thanks,
>>> >>>>>>
>>> >>>>>> Jan
>>> >>>>>>
>>> >>>>>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>>> >>>>>>>> Someone from Flink might correct me if I'm
>>> wrong, but that's my
>>> >>>>>>>> current understanding.
>>> >>>>>>> In essence your description of how exactly-once
>>> works in Flink is
>>> >>>>>>> correct. The general assumption in Flink is that
>>> pipelines must be
>>> >>>>>>> deterministic and thus produce idempotent writes
>>> in the case of
>>> >>>>>>> failures. However, that doesn't mean Beam sinks
>>> can't guarantee a bit
>>> >>>>>>> more with what Flink has to offer.
>>> >>>>>>>
>>> >>>>>>> Luke already mentioned the design discussions
>>> for @RequiresStableInput
>>> >>>>>>> which ensures idempotent writes for
>>> non-deterministic pipelines. This
>>> >>>>>>> is not part of the model but an optional Beam
>>> feature.
>>> >>>>>>>
>>> >>>>>>> We recently implemented support for
>>> @RequiresStableInput in the Flink
>>> >>>>>>> Runner. Reuven mentioned the Flink checkpoint
>>> confirmation, which
>>> >>>>>>> allows us to buffer (and checkpoint) processed
>>> data and only emit it
>>> >>>>>>> once a Flink checkpoint has completed.
>>> >>>>>>>
>>> >>>>>>> Cheers,
>>> >>>>>>> Max
>>> >>>>>>>
>>> >>>>>>> On 21.05.19 16:49, Jan Lukavský wrote:
>>> >>>>>>>> Hi,
>>> >>>>>>>>
>>> >>>>>>>> > Actually, I think it is a larger (open)
>>> question whether exactly
>>> >>>>>>>> once is guaranteed by the model or whether
>>> runners are allowed to
>>> >>>>>>>> relax that. I would think, however, that
>>> sources correctly
>>> >>>>>>>> implemented should be idempotent when run atop
>>> an exactly once
>>> >>>>>>>> infrastructure such as Flink of Dataflow.
>>> >>>>>>>>
>>> >>>>>>>> I would assume, that the model basically
>>> inherits guarantees of
>>> >>>>>>>> underlying infrastructure. Because Flink does
>>> not work as you
>>> >>>>>>>> described (atomic commit of inputs, state and
>>> outputs), but rather a
>>> >>>>>>>> checkpoint mark is flowing through the DAG much
>>> like watermark and on
>>> >>>>>>>> failures operators are restored and data
>>> reprocessed, it (IMHO)
>>> >>>>>>>> implies, that you have exactly once everywhere
>>> in the DAG *but*
>>> >>>>>>>> sinks. That is because sinks cannot be restored
>>> to previous state,
>>> >>>>>>>> instead sinks are supposed to be idempotent in
>>> order for the exactly
>>> >>>>>>>> once to really work (or at least be able to
>>> commit outputs on
>>> >>>>>>>> checkpoint in sink). That implies that if you
>>> don't have sink that is
>>> >>>>>>>> able to commit outputs atomically on
>>> checkpoint, the pipeline
>>> >>>>>>>> execution should be deterministic upon retries,
>>> otherwise shadow
>>> >>>>>>>> writes from failed paths of the pipeline might
>>> appear.
>>> >>>>>>>>
>>> >>>>>>>> Someone from Flink might correct me if I'm
>>> wrong, but that's my
>>> >>>>>>>> current understanding.
>>> >>>>>>>>
>>> >>>>>>>> > Sounds like we should make this clearer.
>>> >>>>>>>>
>>> >>>>>>>> I meant that you are right that we must not in
>>> any thoughts we are
>>> >>>>>>>> having forget that streams are by definition
>>> out-of-order. That is
>>> >>>>>>>> property that we cannot change. But - that
>>> doesn't limit us from
>>> >>>>>>>> creating operator that presents the data to UDF
>>> as if the stream was
>>> >>>>>>>> ideally sorted. It can do that by introducing
>>> latency, of course.
>>> >>>>>>>>
>>> >>>>>>>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>>> >>>>>>>>> Reza: One could provide something like this as
>>> a utility class, but
>>> >>>>>>>>> one downside is that it is not scale
>>> invariant. It requires a tuning
>>> >>>>>>>>> parameter that, if to small, won't mitigate
>>> the problem, but if to
>>> >>>>>>>>> big, greatly increases latency. (Possibly one
>>> could define a dynamic
>>> >>>>>>>>> session-like window to solve this though...)
>>> It also might be harder
>>> >>>>>>>>> for runners that *can* cheaply present stuff
>>> in timestamp order to
>>> >>>>>>>>> optimize. (That and, in practice, our
>>> annotation-style process methods
>>> >>>>>>>>> don't lend themselves to easy composition.) I
>>> think it could work in
>>> >>>>>>>>> specific cases though.
>>> >>>>>>>>>
>>> >>>>>>>>> More inline below.
>>> >>>>>>>>>
>>> >>>>>>>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský
>>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>>> >>>>>>>>>> Hi Robert,
>>> >>>>>>>>>>
>>> >>>>>>>>>> > Beam has an exactly-once model. If the data
>>> was consumed, state
>>> >>>>>>>>>> mutated, and outputs written downstream
>>> (these three are committed
>>> >>>>>>>>>> together atomically) it will not be replayed.
>>> That does not, of
>>> >>>>>>>>>> course,
>>> >>>>>>>>>> solve the non-determanism due to ordering
>>> (including the fact that two
>>> >>>>>>>>>> operations reading the same PCollection may
>>> view different ordering).
>>> >>>>>>>>>>
>>> >>>>>>>>>> I think what you describe is a property of a
>>> runner, not of the model,
>>> >>>>>>>>>> right? I think if I run my pipeline on Flink
>>> I will not get this
>>> >>>>>>>>>> atomicity, because although Flink uses also
>>> exactly-once model if
>>> >>>>>>>>>> might
>>> >>>>>>>>>> write outputs multiple times.
>>> >>>>>>>>> Actually, I think it is a larger (open)
>>> question whether exactly once
>>> >>>>>>>>> is guaranteed by the model or whether runners
>>> are allowed to relax
>>> >>>>>>>>> that. I would think, however, that sources
>>> correctly implemented
>>> >>>>>>>>> should be idempotent when run atop an exactly
>>> once infrastructure such
>>> >>>>>>>>> as Flink of Dataflow.
>>> >>>>>>>>>
>>> >>>>>>>>>> > 1) Is it correct for a (Stateful)DoFn to
>>> assume elements are
>>> >>>>>>>>>> received
>>> >>>>>>>>>> in a specific order? In the current model, it
>>> is not. Being able to
>>> >>>>>>>>>> read, handle, and produced out-of-order data,
>>> including late data,
>>> >>>>>>>>>> is a
>>> >>>>>>>>>> pretty fundamental property of distributed
>>> systems.
>>> >>>>>>>>>>
>>> >>>>>>>>>> Yes, absolutely. The argument here is not
>>> that Stateful ParDo should
>>> >>>>>>>>>> presume to receive elements in any order, but
>>> to _present_ it as
>>> >>>>>>>>>> such to
>>> >>>>>>>>>> the user @ProcessElement function.
>>> >>>>>>>>> Sounds like we should make this clearer.
>>> >>>>>>>>>
>>> >>>>>>>>>> > 2) Given that some operations are easier
>>> (or possibly only
>>> >>>>>>>>>> possible)
>>> >>>>>>>>>> to write when operating on ordered data, and
>>> that different runners
>>> >>>>>>>>>> may
>>> >>>>>>>>>> have (significantly) cheaper ways to provide
>>> this ordering than can be
>>> >>>>>>>>>> done by the user themselves, should we
>>> elevate this to a property of
>>> >>>>>>>>>> (Stateful?)DoFns that the runner can provide?
>>> I think a compelling
>>> >>>>>>>>>> argument can be made here that we should.
>>> >>>>>>>>>>
>>> >>>>>>>>>> +1
>>> >>>>>>>>>>
>>> >>>>>>>>>> Jan
>>> >>>>>>>>>>
>>> >>>>>>>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>>> >>>>>>>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský
>>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>>> >>>>>>>>>>>> > I don't see batch vs. streaming as
>>> part of the model. One
>>> >>>>>>>>>>>> can have
>>> >>>>>>>>>>>> microbatch, or even a runner that
>>> alternates between different
>>> >>>>>>>>>>>> modes.
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> Although I understand motivation of this
>>> statement, this project
>>> >>>>>>>>>>>> name is
>>> >>>>>>>>>>>> "Apache Beam: An advanced unified
>>> programming model". What does the
>>> >>>>>>>>>>>> model unify, if "streaming vs. batch" is
>>> not part of the model?
>>> >>>>>>>>>>> What I mean is that streaming vs. batch is
>>> no longer part of the
>>> >>>>>>>>>>> model
>>> >>>>>>>>>>> (or ideally API), but pushed down to be a
>>> concern of the runner
>>> >>>>>>>>>>> (executor) of the pipeline.
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>
>>> >>>>>>>>>>> On Tue, May 21, 2019 at 10:32 AM Jan
>>> Lukavský <je.ik@seznam.cz <ma...@seznam.cz>>
>>> >>>>>>>>>>> wrote:
>>> >>>>>>>>>>>> Hi Kenn,
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> OK, so if we introduce annotation, we can
>>> have stateful ParDo
>>> >>>>>>>>>>>> with sorting, that would perfectly resolve
>>> my issues. I still
>>> >>>>>>>>>>>> have some doubts, though. Let me explain.
>>> The current behavior of
>>> >>>>>>>>>>>> stateful ParDo has the following properties:
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> a) might fail in batch, although runs
>>> fine in streaming (that
>>> >>>>>>>>>>>> is due to the buffering, and unbounded
>>> lateness in batch, which
>>> >>>>>>>>>>>> was discussed back and forth in this thread)
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> b) might be non deterministic (this is
>>> because the elements
>>> >>>>>>>>>>>> arrive at somewhat random order, and even
>>> if you do the operation
>>> >>>>>>>>>>>> "assign unique ID to elements" this might
>>> produce different
>>> >>>>>>>>>>>> results when run multiple times)
>>> >>>>>>>>>>> PCollections are *explicitly* unordered. Any
>>> operations that
>>> >>>>>>>>>>> assume or
>>> >>>>>>>>>>> depend on a specific ordering for
>>> correctness (or determinism) must
>>> >>>>>>>>>>> provide that ordering themselves (i.e.
>>> tolerate "arbitrary shuffling
>>> >>>>>>>>>>> of inputs"). As you point out, that may be
>>> very expensive if you have
>>> >>>>>>>>>>> very hot keys with very large (unbounded)
>>> timestamp skew.
>>> >>>>>>>>>>>
>>> >>>>>>>>>>> StatefulDoFns are low-level operations that
>>> should be used with care;
>>> >>>>>>>>>>> the simpler windowing model gives
>>> determinism in the face of
>>> >>>>>>>>>>> unordered
>>> >>>>>>>>>>> data (though late data and non-end-of-window
>>> triggering introduces
>>> >>>>>>>>>>> some of the non-determanism back in).
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>> What worries me most is the property b),
>>> because it seems to me
>>> >>>>>>>>>>>> to have serious consequences - not only
>>> that if you run twice
>>> >>>>>>>>>>>> batch pipeline you would get different
>>> results, but even on
>>> >>>>>>>>>>>> streaming, when pipeline fails and gets
>>> restarted from
>>> >>>>>>>>>>>> checkpoint, produced output might differ
>>> from the previous run
>>> >>>>>>>>>>>> and data from the first run might have
>>> already been persisted
>>> >>>>>>>>>>>> into sink. That would create somewhat messy
>>> outputs.
>>> >>>>>>>>>>> Beam has an exactly-once model. If the data
>>> was consumed, state
>>> >>>>>>>>>>> mutated, and outputs written downstream
>>> (these three are committed
>>> >>>>>>>>>>> together atomically) it will not be
>>> replayed. That does not, of
>>> >>>>>>>>>>> course, solve the non-determanism due to
>>> ordering (including the fact
>>> >>>>>>>>>>> that two operations reading the same
>>> PCollection may view different
>>> >>>>>>>>>>> ordering).
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>> These two properties makes me think that
>>> the current
>>> >>>>>>>>>>>> implementation is more of a _special case_
>>> than the general one.
>>> >>>>>>>>>>>> The general one would be that your state
>>> doesn't have the
>>> >>>>>>>>>>>> properties to be able to tolerate buffering
>>> problems and/or
>>> >>>>>>>>>>>> non-determinism. Which is the case where
>>> you need sorting in both
>>> >>>>>>>>>>>> streaming and batch to be part of the model.
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> Let me point out one more analogy - that is
>>> merging vs.
>>> >>>>>>>>>>>> non-merging windows. The general case
>>> (merging windows) implies
>>> >>>>>>>>>>>> sorting by timestamp in both batch case
>>> (explicit) and streaming
>>> >>>>>>>>>>>> (buffering). The special case (non-merging
>>> windows) doesn't rely
>>> >>>>>>>>>>>> on any timestamp ordering, so the sorting
>>> and buffering can be
>>> >>>>>>>>>>>> dropped. The underlying root cause of this
>>> is the same for both
>>> >>>>>>>>>>>> stateful ParDo and windowing (essentially,
>>> assigning window
>>> >>>>>>>>>>>> labels is a stateful operation when
>>> windowing function is merging).
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> The reason for the current behavior of
>>> stateful ParDo seems to be
>>> >>>>>>>>>>>> performance, but is it right to abandon
>>> correctness in favor of
>>> >>>>>>>>>>>> performance? Wouldn't it be more consistent
>>> to have the default
>>> >>>>>>>>>>>> behavior prefer correctness and when you
>>> have the specific
>>> >>>>>>>>>>>> conditions of state function having special
>>> properties, then you
>>> >>>>>>>>>>>> can annotate your DoFn (with something like
>>> >>>>>>>>>>>> @TimeOrderingAgnostic), which would yield a
>>> better performance in
>>> >>>>>>>>>>>> that case?
>>> >>>>>>>>>>> There are two separable questions here.
>>> >>>>>>>>>>>
>>> >>>>>>>>>>> 1) Is it correct for a (Stateful)DoFn to
>>> assume elements are received
>>> >>>>>>>>>>> in a specific order? In the current model,
>>> it is not. Being able to
>>> >>>>>>>>>>> read, handle, and produced out-of-order
>>> data, including late data, is
>>> >>>>>>>>>>> a pretty fundamental property of distributed
>>> systems.
>>> >>>>>>>>>>>
>>> >>>>>>>>>>> 2) Given that some operations are easier (or
>>> possibly only possible)
>>> >>>>>>>>>>> to write when operating on ordered data, and
>>> that different runners
>>> >>>>>>>>>>> may have (significantly) cheaper ways to
>>> provide this ordering than
>>> >>>>>>>>>>> can be done by the user themselves, should
>>> we elevate this to a
>>> >>>>>>>>>>> property of (Stateful?)DoFns that the runner
>>> can provide? I think a
>>> >>>>>>>>>>> compelling argument can be made here that we
>>> should.
>>> >>>>>>>>>>>
>>> >>>>>>>>>>> - Robert
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> Thanks for the nice small example of a
>>> calculation that depends
>>> >>>>>>>>>>>> on order. You are right that many state
>>> machines have this
>>> >>>>>>>>>>>> property. I agree w/ you and Luke that it
>>> is convenient for batch
>>> >>>>>>>>>>>> processing to sort by event timestamp
>>> before running a stateful
>>> >>>>>>>>>>>> ParDo. In streaming you could also
>>> implement "sort by event
>>> >>>>>>>>>>>> timestamp" by buffering until you know all
>>> earlier data will be
>>> >>>>>>>>>>>> dropped - a slack buffer up to allowed
>>> lateness.
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> I do not think that it is OK to sort in
>>> batch and not in
>>> >>>>>>>>>>>> streaming. Many state machines diverge very
>>> rapidly when things
>>> >>>>>>>>>>>> are out of order. So each runner if they
>>> see the
>>> >>>>>>>>>>>> "@OrderByTimestamp" annotation (or
>>> whatever) needs to deliver
>>> >>>>>>>>>>>> sorted data (by some mix of buffering and
>>> dropping), or to reject
>>> >>>>>>>>>>>> the pipeline as unsupported.
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> And also want to say that this is not the
>>> default case - many
>>> >>>>>>>>>>>> uses of state & timers in ParDo yield
>>> different results at the
>>> >>>>>>>>>>>> element level, but the results are
>>> equivalent at in the big
>>> >>>>>>>>>>>> picture. Such as the example of "assign a
>>> unique sequence number
>>> >>>>>>>>>>>> to each element" or "group into batches" it
>>> doesn't matter
>>> >>>>>>>>>>>> exactly what the result is, only that it
>>> meets the spec. And
>>> >>>>>>>>>>>> other cases like user funnels are monotonic
>>> enough that you also
>>> >>>>>>>>>>>> don't actually need sorting.
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> Kenn
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan
>>> Lukavský <je.ik@seznam.cz <ma...@seznam.cz>>
>>> >>>>>>>>>>>> wrote:
>>> >>>>>>>>>>>>> Yes, the problem will arise probably
>>> mostly when you have not
>>> >>>>>>>>>>>>> well distributed keys (or too few keys).
>>> I'm really not sure if
>>> >>>>>>>>>>>>> a pure GBK with a trigger can solve this -
>>> it might help to have
>>> >>>>>>>>>>>>> data driven trigger. There would still be
>>> some doubts, though.
>>> >>>>>>>>>>>>> The main question is still here - people
>>> say, that sorting by
>>> >>>>>>>>>>>>> timestamp before stateful ParDo would be
>>> prohibitively slow, but
>>> >>>>>>>>>>>>> I don't really see why - the sorting is
>>> very probably already
>>> >>>>>>>>>>>>> there. And if not (hash grouping instead
>>> of sorted grouping),
>>> >>>>>>>>>>>>> then the sorting would affect only user
>>> defined StatefulParDos.
>>> >>>>>>>>>>>>>
>>> >>>>>>>>>>>>> This would suggest that the best way out
>>> of this would be really
>>> >>>>>>>>>>>>> to add annotation, so that the author of
>>> the pipeline can decide.
>>> >>>>>>>>>>>>>
>>> >>>>>>>>>>>>> If that would be acceptable I think I can
>>> try to prepare some
>>> >>>>>>>>>>>>> basic functionality, but I'm not sure, if
>>> I would be able to
>>> >>>>>>>>>>>>> cover all runners / sdks.
>>> >>>>>>>>>>>>>
>>> >>>>>>>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>>> >>>>>>>>>>>>>
>>> >>>>>>>>>>>>> It is read all per key and window and not
>>> just read all (this
>>> >>>>>>>>>>>>> still won't scale with hot keys in the
>>> global window). The GBK
>>> >>>>>>>>>>>>> preceding the StatefulParDo will guarantee
>>> that you are
>>> >>>>>>>>>>>>> processing all the values for a specific
>>> key and window at any
>>> >>>>>>>>>>>>> given time. Is there a specific
>>> window/trigger that is missing
>>> >>>>>>>>>>>>> that you feel would remove the need for
>>> you to use StatefulParDo?
>>> >>>>>>>>>>>>>
>>> >>>>>>>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan
>>> Lukavský <je.ik@seznam.cz <ma...@seznam.cz>>
>>> >>>>>>>>>>>>> wrote:
>>> >>>>>>>>>>>>>> Hi Lukasz,
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Today, if you must have a strict order,
>>> you must guarantee
>>> >>>>>>>>>>>>>>> that your StatefulParDo implements the
>>> necessary "buffering &
>>> >>>>>>>>>>>>>>> sorting" into state.
>>> >>>>>>>>>>>>>> Yes, no problem with that. But this whole
>>> discussion started,
>>> >>>>>>>>>>>>>> because *this doesn't work on batch*. You
>>> simply cannot first
>>> >>>>>>>>>>>>>> read everything from distributed storage
>>> and then buffer it all
>>> >>>>>>>>>>>>>> into memory, just to read it again, but
>>> sorted. That will not
>>> >>>>>>>>>>>>>> work. And even if it would, it would be a
>>> terrible waste of
>>> >>>>>>>>>>>>>> resources.
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>> Jan
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan
>>> Lukavský <je.ik@seznam.cz <ma...@seznam.cz>>
>>> >>>>>>>>>>>>>> wrote:
>>> >>>>>>>>>>>>>>> This discussion brings many really
>>> interesting questions for
>>> >>>>>>>>>>>>>>> me. :-)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > I don't see batch vs. streaming as
>>> part of the model. One
>>> >>>>>>>>>>>>>>> can have
>>> >>>>>>>>>>>>>>> microbatch, or even a runner that
>>> alternates between different
>>> >>>>>>>>>>>>>>> modes.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Although I understand motivation of this
>>> statement, this
>>> >>>>>>>>>>>>>>> project name is
>>> >>>>>>>>>>>>>>> "Apache Beam: An advanced unified
>>> programming model". What
>>> >>>>>>>>>>>>>>> does the
>>> >>>>>>>>>>>>>>> model unify, if "streaming vs. batch" is
>>> not part of the model?
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Using microbatching, chaining of batch
>>> jobs, or pure streaming
>>> >>>>>>>>>>>>>>> are
>>> >>>>>>>>>>>>>>> exactly the "runtime
>>> conditions/characteristics" I refer to.
>>> >>>>>>>>>>>>>>> All these
>>> >>>>>>>>>>>>>>> define several runtime parameters, which
>>> in turn define how
>>> >>>>>>>>>>>>>>> well/badly
>>> >>>>>>>>>>>>>>> will the pipeline perform and how many
>>> resources might be
>>> >>>>>>>>>>>>>>> needed. From
>>> >>>>>>>>>>>>>>> my point of view, pure streaming should
>>> be the most resource
>>> >>>>>>>>>>>>>>> demanding
>>> >>>>>>>>>>>>>>> (if not, why bother with batch? why not
>>> run everything in
>>> >>>>>>>>>>>>>>> streaming
>>> >>>>>>>>>>>>>>> only? what will there remain to "unify"?).
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > Fortunately, for batch, only the
>>> state for a single key
>>> >>>>>>>>>>>>>>> needs to be
>>> >>>>>>>>>>>>>>> preserved at a time, rather than the
>>> state for all keys across
>>> >>>>>>>>>>>>>>> the range
>>> >>>>>>>>>>>>>>> of skew. Of course if you have few or
>>> hot keys, one can still
>>> >>>>>>>>>>>>>>> have
>>> >>>>>>>>>>>>>>> issues (and this is not specific to
>>> StatefulDoFns).
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Yes, but here is still the presumption
>>> that my stateful DoFn can
>>> >>>>>>>>>>>>>>> tolerate arbitrary shuffling of inputs.
>>> Let me explain the use
>>> >>>>>>>>>>>>>>> case in
>>> >>>>>>>>>>>>>>> more detail.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Suppose you have input stream consisting
>>> of 1s and 0s (and
>>> >>>>>>>>>>>>>>> some key for
>>> >>>>>>>>>>>>>>> each element, which is irrelevant for
>>> the demonstration). Your
>>> >>>>>>>>>>>>>>> task is
>>> >>>>>>>>>>>>>>> to calculate in running global window
>>> the actual number of
>>> >>>>>>>>>>>>>>> changes
>>> >>>>>>>>>>>>>>> between state 0 and state 1 and vice
>>> versa. When the state
>>> >>>>>>>>>>>>>>> doesn't
>>> >>>>>>>>>>>>>>> change, you don't calculate anything. If
>>> input (for given key)
>>> >>>>>>>>>>>>>>> would be
>>> >>>>>>>>>>>>>>> (tN denotes timestamp N):
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t1: 1
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t2: 0
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t3: 0
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t4: 1
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t5: 1
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t6: 0
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> then the output should yield (supposing
>>> that default state is
>>> >>>>>>>>>>>>>>> zero):
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t1: (one: 1, zero: 0)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t2: (one: 1, zero: 1)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t3: (one: 1, zero: 1)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t4: (one: 2, zero: 1)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t5: (one: 2, zero: 1)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t6: (one: 2, zero: 2)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> How would you implement this in current
>>> Beam semantics?
>>> >>>>>>>>>>>>>> I think your saying here that I know that
>>> my input is ordered
>>> >>>>>>>>>>>>>> in a specific way and since I assume the
>>> order when writing my
>>> >>>>>>>>>>>>>> pipeline I can perform this optimization.
>>> But there is nothing
>>> >>>>>>>>>>>>>> preventing a runner from noticing that
>>> your processing in the
>>> >>>>>>>>>>>>>> global window with a specific type of
>>> trigger and re-ordering
>>> >>>>>>>>>>>>>> your inputs/processing to get better
>>> performance (since you
>>> >>>>>>>>>>>>>> can't use an AfterWatermark trigger for
>>> your pipeline in
>>> >>>>>>>>>>>>>> streaming for the GlobalWindow).
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>> Today, if you must have a strict order,
>>> you must guarantee that
>>> >>>>>>>>>>>>>> your StatefulParDo implements the
>>> necessary "buffering &
>>> >>>>>>>>>>>>>> sorting" into state. I can see why you
>>> would want an annotation
>>> >>>>>>>>>>>>>> that says I must have timestamp ordered
>>> elements, since it
>>> >>>>>>>>>>>>>> makes writing certain StatefulParDos much
>>> easier. StatefulParDo
>>> >>>>>>>>>>>>>> is a low-level function, it really is the
>>> "here you go and do
>>> >>>>>>>>>>>>>> whatever you need to but here be dragons"
>>> function while
>>> >>>>>>>>>>>>>> windowing and triggering is meant to keep
>>> many people from
>>> >>>>>>>>>>>>>> writing StatefulParDo in the first place.
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > Pipelines that fail in the "worst
>>> case" batch scenario
>>> >>>>>>>>>>>>>>> are likely to
>>> >>>>>>>>>>>>>>> degrade poorly (possibly
>>> catastrophically) when the watermark
>>> >>>>>>>>>>>>>>> falls
>>> >>>>>>>>>>>>>>> behind in streaming mode as well.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> But the worst case is defined by input
>>> of size (available
>>> >>>>>>>>>>>>>>> resources +
>>> >>>>>>>>>>>>>>> single byte) -> pipeline fail. Although
>>> it could have
>>> >>>>>>>>>>>>>>> finished, given
>>> >>>>>>>>>>>>>>> the right conditions.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > This might be reasonable,
>>> implemented by default by
>>> >>>>>>>>>>>>>>> buffering
>>> >>>>>>>>>>>>>>> everything and releasing elements as the
>>> watermark (+lateness)
>>> >>>>>>>>>>>>>>> advances,
>>> >>>>>>>>>>>>>>> but would likely lead to inefficient
>>> (though *maybe* easier to
>>> >>>>>>>>>>>>>>> reason
>>> >>>>>>>>>>>>>>> about) code.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Sure, the pipeline will be less
>>> efficient, because it would
>>> >>>>>>>>>>>>>>> have to
>>> >>>>>>>>>>>>>>> buffer and sort the inputs. But at least
>>> it will produce
>>> >>>>>>>>>>>>>>> correct results
>>> >>>>>>>>>>>>>>> in cases where updates to state are
>>> order-sensitive.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > Would it be roughly equivalent to
>>> GBK + FlatMap(lambda
>>> >>>>>>>>>>>>>>> (key, values):
>>> >>>>>>>>>>>>>>> [(key, value) for value in values])?
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> I'd say roughly yes, but difference
>>> would be in the trigger.
>>> >>>>>>>>>>>>>>> The trigger
>>> >>>>>>>>>>>>>>> should ideally fire as soon as watermark
>>> (+lateness) crosses
>>> >>>>>>>>>>>>>>> element
>>> >>>>>>>>>>>>>>> with lowest timestamp in the buffer.
>>> Although this could be
>>> >>>>>>>>>>>>>>> somehow
>>> >>>>>>>>>>>>>>> emulated by fixed trigger each X millis.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > Or is the underlying desire just
>>> to be able to hint to
>>> >>>>>>>>>>>>>>> the runner
>>> >>>>>>>>>>>>>>> that the code may perform better (e.g.
>>> require less resources)
>>> >>>>>>>>>>>>>>> as skew
>>> >>>>>>>>>>>>>>> is reduced (and hence to order by
>>> timestamp iff it's cheap)?
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> No, the sorting would have to be done in
>>> streaming case as
>>> >>>>>>>>>>>>>>> well. That is
>>> >>>>>>>>>>>>>>> an imperative of the unified model. I
>>> think it is possible to
>>> >>>>>>>>>>>>>>> sort by
>>> >>>>>>>>>>>>>>> timestamp only in batch case (and do it
>>> for *all* batch
>>> >>>>>>>>>>>>>>> stateful pardos
>>> >>>>>>>>>>>>>>> without annotation), or introduce
>>> annotation, but then make
>>> >>>>>>>>>>>>>>> the same
>>> >>>>>>>>>>>>>>> guarantees for streaming case as well.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Jan
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>>> >>>>>>>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan
>>> Lukavský
>>> >>>>>>>>>>>>>>>> <je.ik@seznam.cz
>>> <ma...@seznam.cz>> wrote:
>>> >>>>>>>>>>>>>>>>> Hi Robert,
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> yes, I think you rephrased my point -
>>> although no *explicit*
>>> >>>>>>>>>>>>>>>>> guarantees
>>> >>>>>>>>>>>>>>>>> of ordering are given in either mode,
>>> there is *implicit*
>>> >>>>>>>>>>>>>>>>> ordering in
>>> >>>>>>>>>>>>>>>>> streaming case that is due to nature
>>> of the processing - the
>>> >>>>>>>>>>>>>>>>> difference
>>> >>>>>>>>>>>>>>>>> between watermark and timestamp of
>>> elements flowing through
>>> >>>>>>>>>>>>>>>>> the pipeline
>>> >>>>>>>>>>>>>>>>> are generally low (too high difference
>>> leads to the
>>> >>>>>>>>>>>>>>>>> overbuffering
>>> >>>>>>>>>>>>>>>>> problem), but there is no such bound
>>> on batch.
>>> >>>>>>>>>>>>>>>> Fortunately, for batch, only the state
>>> for a single key needs
>>> >>>>>>>>>>>>>>>> to be
>>> >>>>>>>>>>>>>>>> preserved at a time, rather than the
>>> state for all keys
>>> >>>>>>>>>>>>>>>> across the
>>> >>>>>>>>>>>>>>>> range of skew. Of course if you have
>>> few or hot keys, one can
>>> >>>>>>>>>>>>>>>> still
>>> >>>>>>>>>>>>>>>> have issues (and this is not specific
>>> to StatefulDoFns).
>>> >>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> As a result, I see a few possible
>>> solutions:
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> - the best and most natural seems
>>> to be extension of
>>> >>>>>>>>>>>>>>>>> the model, so
>>> >>>>>>>>>>>>>>>>> that it defines batch as not only
>>> "streaming pipeline
>>> >>>>>>>>>>>>>>>>> executed in batch
>>> >>>>>>>>>>>>>>>>> fashion", but "pipeline with at least
>>> as good runtime
>>> >>>>>>>>>>>>>>>>> characteristics as
>>> >>>>>>>>>>>>>>>>> in streaming case, executed in batch
>>> fashion", I really
>>> >>>>>>>>>>>>>>>>> don't think that
>>> >>>>>>>>>>>>>>>>> there are any conflicts with the
>>> current model, or that this
>>> >>>>>>>>>>>>>>>>> could
>>> >>>>>>>>>>>>>>>>> affect performance, because the
>>> required sorting (as pointed by
>>> >>>>>>>>>>>>>>>>> Aljoscha) is very probably already
>>> done during translation
>>> >>>>>>>>>>>>>>>>> of stateful
>>> >>>>>>>>>>>>>>>>> pardos. Also note that this definition
>>> only affects user
>>> >>>>>>>>>>>>>>>>> defined
>>> >>>>>>>>>>>>>>>>> stateful pardos
>>> >>>>>>>>>>>>>>>> I don't see batch vs. streaming as part
>>> of the model. One can
>>> >>>>>>>>>>>>>>>> have
>>> >>>>>>>>>>>>>>>> microbatch, or even a runner that
>>> alternates between
>>> >>>>>>>>>>>>>>>> different modes.
>>> >>>>>>>>>>>>>>>> The model describes what the valid
>>> outputs are given a
>>> >>>>>>>>>>>>>>>> (sometimes
>>> >>>>>>>>>>>>>>>> partial) set of inputs. It becomes
>>> really hard to define
>>> >>>>>>>>>>>>>>>> things like
>>> >>>>>>>>>>>>>>>> "as good runtime characteristics." Once
>>> you allow any
>>> >>>>>>>>>>>>>>>> out-of-orderedness, it is not very
>>> feasible to try and define
>>> >>>>>>>>>>>>>>>> (and
>>> >>>>>>>>>>>>>>>> more cheaply implement) a "upper bound"
>>> of acceptable
>>> >>>>>>>>>>>>>>>> out-of-orderedness.
>>> >>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>> Pipelines that fail in the "worst case"
>>> batch scenario are
>>> >>>>>>>>>>>>>>>> likely to
>>> >>>>>>>>>>>>>>>> degrade poorly (possibly
>>> catastrophically) when the watermark
>>> >>>>>>>>>>>>>>>> falls
>>> >>>>>>>>>>>>>>>> behind in streaming mode as well.
>>> >>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> - another option would be to
>>> introduce annotation for
>>> >>>>>>>>>>>>>>>>> DoFns (e.g.
>>> >>>>>>>>>>>>>>>>> @RequiresStableTimeCharacteristics),
>>> which would result in
>>> >>>>>>>>>>>>>>>>> the sorting
>>> >>>>>>>>>>>>>>>>> in batch case - but - this extension
>>> would have to ensure
>>> >>>>>>>>>>>>>>>>> the sorting in
>>> >>>>>>>>>>>>>>>>> streaming mode also - it would require
>>> definition of allowed
>>> >>>>>>>>>>>>>>>>> lateness,
>>> >>>>>>>>>>>>>>>>> and triggger (essentially similar to
>>> window)
>>> >>>>>>>>>>>>>>>> This might be reasonable, implemented
>>> by default by buffering
>>> >>>>>>>>>>>>>>>> everything and releasing elements as
>>> the watermark (+lateness)
>>> >>>>>>>>>>>>>>>> advances, but would likely lead to
>>> inefficient (though
>>> >>>>>>>>>>>>>>>> *maybe* easier
>>> >>>>>>>>>>>>>>>> to reason about) code. Not sure about
>>> the semantics of
>>> >>>>>>>>>>>>>>>> triggering
>>> >>>>>>>>>>>>>>>> here, especially data-driven triggers.
>>> Would it be roughly
>>> >>>>>>>>>>>>>>>> equivalent
>>> >>>>>>>>>>>>>>>> to GBK + FlatMap(lambda (key, values):
>>> [(key, value) for
>>> >>>>>>>>>>>>>>>> value in
>>> >>>>>>>>>>>>>>>> values])?
>>> >>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>> Or is the underlying desire just to be
>>> able to hint to the
>>> >>>>>>>>>>>>>>>> runner that
>>> >>>>>>>>>>>>>>>> the code may perform better (e.g.
>>> require less resources) as
>>> >>>>>>>>>>>>>>>> skew is
>>> >>>>>>>>>>>>>>>> reduced (and hence to order by
>>> timestamp iff it's cheap)?
>>> >>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> - last option would be to introduce
>>> these "higher order
>>> >>>>>>>>>>>>>>>>> guarantees" in
>>> >>>>>>>>>>>>>>>>> some extension DSL (e.g. Euphoria),
>>> but that seems to be the
>>> >>>>>>>>>>>>>>>>> worst
>>> >>>>>>>>>>>>>>>>> option to me
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> I see the first two options quite
>>> equally good, although the
>>> >>>>>>>>>>>>>>>>> letter one
>>> >>>>>>>>>>>>>>>>> is probably more time consuming to
>>> implement. But it would
>>> >>>>>>>>>>>>>>>>> bring
>>> >>>>>>>>>>>>>>>>> additional feature to streaming case
>>> as well.
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> Thanks for any thoughts.
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> Jan
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw
>>> wrote:
>>> >>>>>>>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan
>>> Lukavský
>>> >>>>>>>>>>>>>>>>>> <je.ik@seznam.cz
>>> <ma...@seznam.cz>> wrote:
>>> >>>>>>>>>>>>>>>>>>> Hi Reuven,
>>> >>>>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work
>>> just fine in batch
>>> >>>>>>>>>>>>>>>>>>>> runners.
>>> >>>>>>>>>>>>>>>>>>> Stateful ParDo works in batch as
>>> far, as the logic inside
>>> >>>>>>>>>>>>>>>>>>> the state works for absolutely
>>> unbounded out-of-orderness
>>> >>>>>>>>>>>>>>>>>>> of elements. That basically
>>> (practically) can work only
>>> >>>>>>>>>>>>>>>>>>> for cases, where the order of input
>>> elements doesn't
>>> >>>>>>>>>>>>>>>>>>> matter. But, "state" can refer to
>>> "state machine", and any
>>> >>>>>>>>>>>>>>>>>>> time you have a state machine
>>> involved, then the ordering
>>> >>>>>>>>>>>>>>>>>>> of elements would matter.
>>> >>>>>>>>>>>>>>>>>> No guarantees on order are provided
>>> in *either* streaming
>>> >>>>>>>>>>>>>>>>>> or batch
>>> >>>>>>>>>>>>>>>>>> mode by the model. However, it is the
>>> case that in order to
>>> >>>>>>>>>>>>>>>>>> make
>>> >>>>>>>>>>>>>>>>>> forward progress most streaming
>>> runners attempt to limit
>>> >>>>>>>>>>>>>>>>>> the amount of
>>> >>>>>>>>>>>>>>>>>> out-of-orderedness of elements (in
>>> terms of event time vs.
>>> >>>>>>>>>>>>>>>>>> processing
>>> >>>>>>>>>>>>>>>>>> time) to make forward progress, which
>>> in turn could help
>>> >>>>>>>>>>>>>>>>>> cap the
>>> >>>>>>>>>>>>>>>>>> amount of state that must be held
>>> concurrently, whereas a
>>> >>>>>>>>>>>>>>>>>> batch runner
>>> >>>>>>>>>>>>>>>>>> may not allow any state to be safely
>>> discarded until the whole
>>> >>>>>>>>>>>>>>>>>> timeline from infinite past to
>>> infinite future has been
>>> >>>>>>>>>>>>>>>>>> observed.
>>> >>>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>>> Also, as pointed out, state is not
>>> preserved "batch to
>>> >>>>>>>>>>>>>>>>>> batch" in batch mode.
>>> >>>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM
>>> Maximilian Michels
>>> >>>>>>>>>>>>>>>>>> <mxm@apache.org
>>> <ma...@apache.org>> wrote:
>>> >>>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>>>>> batch semantics and streaming
>>> semantics differs only
>>> >>>>>>>>>>>>>>>>>>>> in that I can have GlobalWindow
>>> with default trigger on
>>> >>>>>>>>>>>>>>>>>>>> batch and cannot on stream
>>> >>>>>>>>>>>>>>>>>>> You can have a GlobalWindow in
>>> streaming with a default
>>> >>>>>>>>>>>>>>>>>>> trigger. You
>>> >>>>>>>>>>>>>>>>>>> could define additional triggers
>>> that do early firings.
>>> >>>>>>>>>>>>>>>>>>> And you could
>>> >>>>>>>>>>>>>>>>>>> even trigger the global window by
>>> advancing the watermark
>>> >>>>>>>>>>>>>>>>>>> to +inf.
>>> >>>>>>>>>>>>>>>>>> IIRC, as a pragmatic note, we
>>> prohibited global window with
>>> >>>>>>>>>>>>>>>>>> default
>>> >>>>>>>>>>>>>>>>>> trigger on unbounded PCollections in
>>> the SDK because this
>>> >>>>>>>>>>>>>>>>>> is more
>>> >>>>>>>>>>>>>>>>>> likely to be user error than an
>>> actual desire to have no
>>> >>>>>>>>>>>>>>>>>> output until
>>> >>>>>>>>>>>>>>>>>> drain. But it's semantically valid in
>>> the model.
>>>
Re: Definition of Unified model
Posted by Lukasz Cwik <lc...@google.com>.
Expanding the dimensionality could be the basis for loops within the graph
since loops could be modeled as (time, loop iteration #, nested loop
iteration #, nested nested loop iteration #, ...)
On Tue, May 28, 2019 at 12:10 PM Jan Lukavský <je...@seznam.cz> wrote:
> Could this be solved by "expanding the dimensionality" of time field? What
> I mean by that - if input element to to FlatMap has sequence number T, then
> the (stateless) FlatMap knows the ordering of output elements, right? If it
> would expand the field by which it will next sort the elements to (X, 1),
> (X, 2), ... (X, N), then it would be possible to sort the elements back
> later. There seems to be no need for state to achieve that, or?
>
> Jan
> On 5/28/19 6:52 PM, Reuven Lax wrote:
>
> A slightly larger concern: it also will force users to create stateful
> DoFns everywhere to generate these sequence numbers. If I have a ParDo that
> is not a simple 1:1 transform (i.e. not MapElements), then the ParDo will
> need to generate its own sequence numbers for ordering, and the only safe
> way to do so is to use a stateful DoFn. This turns what used to be a simple
> in-memory DoFn into one that has to access state. Also I believe many
> runners will not fuse stateful DoFns. While none of this poses a problem
> for the model, it could make ordering extremely expensive to achieve.
>
> Reuven
>
> On Tue, May 28, 2019 at 6:09 AM Jan Lukavský <je...@seznam.cz> wrote:
>
>> Hi Reuven,
>>
>> > It also gets awkward with Flatten - the sequence number is no longer
>> enough, you must also encode which side of the flatten each element came
>> from.
>>
>> That is a generic need. Even if you read data from Kafka, the offsets are
>> comparable only inside single partition. So, for Kafka to work as a FIFO
>> for ordering, elements with same key have to be pushed to the same
>> partition (otherwise Kafka cannot act as FIFO, because different partitions
>> can be handled by different brokers, which means different observers and
>> they therefore might not agree on the order of events). So if we want to
>> emulate FIFO per key, then the sequence IDs have also be per key.
>> On 5/28/19 2:33 PM, Reuven Lax wrote:
>>
>> Sequence metadata does have the disadvantage that users can no longer use
>> the types coming from the source. You must create a new type that contains
>> a sequence number (unless Beam provides this). It also gets awkward with
>> Flatten - the sequence number is no longer enough, you must also encode
>> which side of the flatten each element came from.
>>
>> On Tue, May 28, 2019 at 3:18 AM Jan Lukavský <je...@seznam.cz> wrote:
>>
>>> As I understood it, Kenn was supporting the idea that sequence metadata
>>> is preferable over FIFO. I was trying to point out, that it even should
>>> provide the same functionally as FIFO, plus one important more -
>>> reproducibility and ability to being persisted and reused the same way
>>> in batch and streaming.
>>>
>>> There is no doubt, that sequence metadata can be stored in every
>>> storage. But, regarding some implicit ordering that sources might have -
>>> yes, of course, data written into HDFS or Cloud Storage has ordering,
>>> but only partial - inside some bulk (e.g. file) and the ordering is not
>>> defined correctly on boundaries of these bulks (between files). That is
>>> why I'd say, that ordering of sources is relevant only for
>>> (partitioned!) streaming sources and generally always reduces to
>>> sequence metadata (e.g. offsets).
>>>
>>> Jan
>>>
>>> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
>>> > Huge +1 to all Kenn said.
>>> >
>>> > Jan, batch sources can have orderings too, just like Kafka. I think
>>> > it's reasonable (for both batch and streaming) that if a source has an
>>> > ordering that is an important part of the data, it should preserve
>>> > this ordering into the data itself (e.g. as sequence numbers, offsets,
>>> > etc.)
>>> >
>>> > On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org>
>>> wrote:
>>> >> I strongly prefer explicit sequence metadata over FIFO requirements,
>>> because:
>>> >>
>>> >> - FIFO is complex to specify: for example Dataflow has "per stage
>>> key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage"
>>> is not a portable concept, nor even guaranteed to remain a Dataflow concept)
>>> >> - complex specifications are by definition poor usability (if
>>> necessary, then it is what it is)
>>> >> - overly restricts the runner, reduces parallelism, for example any
>>> non-stateful ParDo has per-element parallelism, not per "key"
>>> >> - another perspective on that: FIFO makes everyone pay rather than
>>> just the transform that requires exactly sequencing
>>> >> - previous implementation details like reshuffles become part of
>>> the model
>>> >> - I'm not even convinced the use cases involved are addressed by
>>> some careful FIFO restrictions; many sinks re-key and they would all have
>>> to become aware of how keying of a sequence of "stages" affects the
>>> end-to-end FIFO
>>> >>
>>> >> A noop becoming a non-noop is essentially the mathematical definition
>>> of moving from higher-level to lower-level abstraction.
>>> >>
>>> >> So this strikes at the core question of what level of abstraction
>>> Beam aims to represent. Lower-level means there are fewer possible
>>> implementations and it is more tied to the underlying architecture, and
>>> anything not near-exact match pays a huge penalty. Higher-level means there
>>> are more implementations possible with different tradeoffs, though they may
>>> all pay a minor penalty.
>>> >>
>>> >> I could be convinced to change my mind, but it needs some extensive
>>> design, examples, etc. I think it is probably about the most consequential
>>> design decision in the whole Beam model, around the same level as the
>>> decision to use ParDo and GBK as the primitives IMO.
>>> >>
>>> >> Kenn
>>> >>
>>> >> On Thu, May 23, 2019 at 10:17 AM Reuven Lax <re...@google.com> wrote:
>>> >>> Not really. I'm suggesting that some variant of FIFO ordering is
>>> necessary, which requires either runners natively support FIFO ordering or
>>> transforms adding some extra sequence number to each record to sort by.
>>> >>>
>>> >>> I still think your proposal is very useful by the way. I'm merely
>>> pointing out that to solve the state-machine problem we probably need
>>> something more.
>>> >>>
>>> >>> Reuven
>>> >>>
>>> >>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je...@seznam.cz>
>>> wrote:
>>> >>>> Hi,
>>> >>>> yes. It seems that ordering by user supplied UDF makes sense and I
>>> will update the design proposal accordingly.
>>> >>>> Would that solve the issues you mention?
>>> >>>> Jan
>>> >>>> ---------- Původní e-mail ----------
>>> >>>> Od: Reuven Lax <re...@google.com>
>>> >>>> Komu: dev <de...@beam.apache.org>
>>> >>>> Datum: 23. 5. 2019 18:44:38
>>> >>>> Předmět: Re: Definition of Unified model
>>> >>>>
>>> >>>> I'm simply saying that timestamp ordering is insufficient for state
>>> machines. I wasn't proposing Kafka as a solution - that was simply an
>>> example of how people solve this problem in other scenarios.
>>> >>>>
>>> >>>> BTW another example of ordering: Imagine today that you have a
>>> triggered Sum aggregation writing out to a key-value sink. In theory we
>>> provide no ordering, so the sink might write the triggered sums in the
>>> wrong order, ending up with an incorrect value in the sink. In this case
>>> you probably want values ordered by trigger pane index.
>>> >>>>
>>> >>>> Reuven
>>> >>>>
>>> >>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz>
>>> wrote:
>>> >>>>
>>> >>>> Hi Reuven,
>>> >>>> I share the view point of Robert. I think the isuue you refer to is
>>> not in reality related to timestamps, but to the fact, that ordering of
>>> events in time is observer dependent (either caused by relativity, or time
>>> skew, essentially this has the same consequences). And the resolution in
>>> fact isn't Kafka, but generally an authoritative observer, that tells you
>>> "I saw the events in this order". And you either have one (and have the
>>> outcome of his observation persisted in the data - e.g. as offset in Kafka
>>> partition), then you should be able to use it (maybe that suggests afterall
>>> that sorting by some user supplied UDF might make sense), or do not have
>>> it, and then any interpretation of the data seems to be equally valid.
>>> Although determinism is fine, of course.
>>> >>>> Jan
>>> >>>> ---------- Původní e-mail ----------
>>> >>>> Od: Reuven Lax <re...@google.com>
>>> >>>> Komu: dev <de...@beam.apache.org>
>>> >>>> Datum: 23. 5. 2019 17:39:12
>>> >>>> Předmět: Re: Definition of Unified model
>>> >>>>
>>> >>>> So an example would be elements of type "startUserSession" and
>>> "endUserSession" (website sessions, not Beam sessions). Logically you may
>>> need to process them in the correct order if you have any sort of
>>> state-machine logic. However timestamp ordering is never guaranteed to
>>> match the logical ordering. Not only might you have several elements with
>>> the same timestamp, but in reality time skew across backend servers can
>>> cause the events to have timestamps in reverse order of the actual
>>> causality order.
>>> >>>>
>>> >>>> People do solve this problem today though. Publish the events to
>>> Kafka, making sure that events for the same user end up in the same Kafka
>>> partition. This ensures that the events appear in the Kafka partitions in
>>> causality order, even if the timestamp order doesn't match. The your Kafka
>>> subscriber simply process the elements in each partition in order.
>>> >>>>
>>> >>>> I think the ability to impose FIFO causality ordering is what's
>>> needed for any state-machine work. Timestamp ordering has advantages
>>> (though often I think the advantage is in state), but does not solve this
>>> problem.
>>> >>>>
>>> >>>> Reuven
>>> >>>>
>>> >>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <
>>> robertwb@google.com> wrote:
>>> >>>>
>>> >>>> Good point.
>>> >>>>
>>> >>>> The "implementation-specific" way I would do this is
>>> >>>> window-by-instant, followed by a DoFn that gets all the elements
>>> with
>>> >>>> the same timestamp and sorts/acts accordingly, but this counts on
>>> the
>>> >>>> runner producing windows in timestamp order (likely?) and also the
>>> >>>> subsequent DoFn getting them in this order (also likely, due to
>>> >>>> fusion).
>>> >>>>
>>> >>>> One could make the argument that, though it does not provide
>>> >>>> deterministic behavior, getting elements of the same timestamp in
>>> >>>> different orders should produce equally valid interpretations of the
>>> >>>> data. (After all, due to relatively, timestamps are not technically
>>> >>>> well ordered across space.) I can see how data-dependent tiebreakers
>>> >>>> could be useful, or promises of preservation of order between
>>> >>>> operations.
>>> >>>>
>>> >>>> - Robert
>>> >>>>
>>> >>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com>
>>> wrote:
>>> >>>>> So Jan's example of state machines is quite a valid use case for
>>> ordering. However in my experience, timestamp ordering is insufficient for
>>> state machines. Elements that cause state transitions might come in with
>>> the exact same timestamp, yet still have a necessary ordering. Especially
>>> given Beam's decision to have milliseconds timestamps this is possible, but
>>> even at microsecond or nanosecond precision this can happen at scale. To
>>> handle state machines you usually need some sort of FIFO ordering along
>>> with an ordered sources, such as Kafka, not timestamp ordering.
>>> >>>>>
>>> >>>>> Reuven
>>> >>>>>
>>> >>>>> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz>
>>> wrote:
>>> >>>>>> Hi all,
>>> >>>>>>
>>> >>>>>> thanks everyone for this discussion. I think I have gathered
>>> enough
>>> >>>>>> feedback to be able to put down a proposition for changes, which
>>> I will
>>> >>>>>> do and send to this list for further discussion. There are still
>>> doubts
>>> >>>>>> remaining the non-determinism and it's relation to outputs
>>> stability vs.
>>> >>>>>> latency. But I will try to clarify all this in the design
>>> document.
>>> >>>>>>
>>> >>>>>> Thanks,
>>> >>>>>>
>>> >>>>>> Jan
>>> >>>>>>
>>> >>>>>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>>> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
>>> >>>>>>>> current understanding.
>>> >>>>>>> In essence your description of how exactly-once works in Flink is
>>> >>>>>>> correct. The general assumption in Flink is that pipelines must
>>> be
>>> >>>>>>> deterministic and thus produce idempotent writes in the case of
>>> >>>>>>> failures. However, that doesn't mean Beam sinks can't guarantee
>>> a bit
>>> >>>>>>> more with what Flink has to offer.
>>> >>>>>>>
>>> >>>>>>> Luke already mentioned the design discussions for
>>> @RequiresStableInput
>>> >>>>>>> which ensures idempotent writes for non-deterministic pipelines.
>>> This
>>> >>>>>>> is not part of the model but an optional Beam feature.
>>> >>>>>>>
>>> >>>>>>> We recently implemented support for @RequiresStableInput in the
>>> Flink
>>> >>>>>>> Runner. Reuven mentioned the Flink checkpoint confirmation, which
>>> >>>>>>> allows us to buffer (and checkpoint) processed data and only
>>> emit it
>>> >>>>>>> once a Flink checkpoint has completed.
>>> >>>>>>>
>>> >>>>>>> Cheers,
>>> >>>>>>> Max
>>> >>>>>>>
>>> >>>>>>> On 21.05.19 16:49, Jan Lukavský wrote:
>>> >>>>>>>> Hi,
>>> >>>>>>>>
>>> >>>>>>>> > Actually, I think it is a larger (open) question whether
>>> exactly
>>> >>>>>>>> once is guaranteed by the model or whether runners are allowed
>>> to
>>> >>>>>>>> relax that. I would think, however, that sources correctly
>>> >>>>>>>> implemented should be idempotent when run atop an exactly once
>>> >>>>>>>> infrastructure such as Flink of Dataflow.
>>> >>>>>>>>
>>> >>>>>>>> I would assume, that the model basically inherits guarantees of
>>> >>>>>>>> underlying infrastructure. Because Flink does not work as you
>>> >>>>>>>> described (atomic commit of inputs, state and outputs), but
>>> rather a
>>> >>>>>>>> checkpoint mark is flowing through the DAG much like watermark
>>> and on
>>> >>>>>>>> failures operators are restored and data reprocessed, it (IMHO)
>>> >>>>>>>> implies, that you have exactly once everywhere in the DAG *but*
>>> >>>>>>>> sinks. That is because sinks cannot be restored to previous
>>> state,
>>> >>>>>>>> instead sinks are supposed to be idempotent in order for the
>>> exactly
>>> >>>>>>>> once to really work (or at least be able to commit outputs on
>>> >>>>>>>> checkpoint in sink). That implies that if you don't have sink
>>> that is
>>> >>>>>>>> able to commit outputs atomically on checkpoint, the pipeline
>>> >>>>>>>> execution should be deterministic upon retries, otherwise shadow
>>> >>>>>>>> writes from failed paths of the pipeline might appear.
>>> >>>>>>>>
>>> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
>>> >>>>>>>> current understanding.
>>> >>>>>>>>
>>> >>>>>>>> > Sounds like we should make this clearer.
>>> >>>>>>>>
>>> >>>>>>>> I meant that you are right that we must not in any thoughts we
>>> are
>>> >>>>>>>> having forget that streams are by definition out-of-order. That
>>> is
>>> >>>>>>>> property that we cannot change. But - that doesn't limit us from
>>> >>>>>>>> creating operator that presents the data to UDF as if the
>>> stream was
>>> >>>>>>>> ideally sorted. It can do that by introducing latency, of
>>> course.
>>> >>>>>>>>
>>> >>>>>>>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>>> >>>>>>>>> Reza: One could provide something like this as a utility
>>> class, but
>>> >>>>>>>>> one downside is that it is not scale invariant. It requires a
>>> tuning
>>> >>>>>>>>> parameter that, if to small, won't mitigate the problem, but
>>> if to
>>> >>>>>>>>> big, greatly increases latency. (Possibly one could define a
>>> dynamic
>>> >>>>>>>>> session-like window to solve this though...) It also might be
>>> harder
>>> >>>>>>>>> for runners that *can* cheaply present stuff in timestamp
>>> order to
>>> >>>>>>>>> optimize. (That and, in practice, our annotation-style process
>>> methods
>>> >>>>>>>>> don't lend themselves to easy composition.) I think it could
>>> work in
>>> >>>>>>>>> specific cases though.
>>> >>>>>>>>>
>>> >>>>>>>>> More inline below.
>>> >>>>>>>>>
>>> >>>>>>>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz>
>>> wrote:
>>> >>>>>>>>>> Hi Robert,
>>> >>>>>>>>>>
>>> >>>>>>>>>> > Beam has an exactly-once model. If the data was
>>> consumed, state
>>> >>>>>>>>>> mutated, and outputs written downstream (these three are
>>> committed
>>> >>>>>>>>>> together atomically) it will not be replayed. That does not,
>>> of
>>> >>>>>>>>>> course,
>>> >>>>>>>>>> solve the non-determanism due to ordering (including the fact
>>> that two
>>> >>>>>>>>>> operations reading the same PCollection may view different
>>> ordering).
>>> >>>>>>>>>>
>>> >>>>>>>>>> I think what you describe is a property of a runner, not of
>>> the model,
>>> >>>>>>>>>> right? I think if I run my pipeline on Flink I will not get
>>> this
>>> >>>>>>>>>> atomicity, because although Flink uses also exactly-once
>>> model if
>>> >>>>>>>>>> might
>>> >>>>>>>>>> write outputs multiple times.
>>> >>>>>>>>> Actually, I think it is a larger (open) question whether
>>> exactly once
>>> >>>>>>>>> is guaranteed by the model or whether runners are allowed to
>>> relax
>>> >>>>>>>>> that. I would think, however, that sources correctly
>>> implemented
>>> >>>>>>>>> should be idempotent when run atop an exactly once
>>> infrastructure such
>>> >>>>>>>>> as Flink of Dataflow.
>>> >>>>>>>>>
>>> >>>>>>>>>> > 1) Is it correct for a (Stateful)DoFn to assume elements
>>> are
>>> >>>>>>>>>> received
>>> >>>>>>>>>> in a specific order? In the current model, it is not. Being
>>> able to
>>> >>>>>>>>>> read, handle, and produced out-of-order data, including late
>>> data,
>>> >>>>>>>>>> is a
>>> >>>>>>>>>> pretty fundamental property of distributed systems.
>>> >>>>>>>>>>
>>> >>>>>>>>>> Yes, absolutely. The argument here is not that Stateful ParDo
>>> should
>>> >>>>>>>>>> presume to receive elements in any order, but to _present_ it
>>> as
>>> >>>>>>>>>> such to
>>> >>>>>>>>>> the user @ProcessElement function.
>>> >>>>>>>>> Sounds like we should make this clearer.
>>> >>>>>>>>>
>>> >>>>>>>>>> > 2) Given that some operations are easier (or possibly
>>> only
>>> >>>>>>>>>> possible)
>>> >>>>>>>>>> to write when operating on ordered data, and that different
>>> runners
>>> >>>>>>>>>> may
>>> >>>>>>>>>> have (significantly) cheaper ways to provide this ordering
>>> than can be
>>> >>>>>>>>>> done by the user themselves, should we elevate this to a
>>> property of
>>> >>>>>>>>>> (Stateful?)DoFns that the runner can provide? I think a
>>> compelling
>>> >>>>>>>>>> argument can be made here that we should.
>>> >>>>>>>>>>
>>> >>>>>>>>>> +1
>>> >>>>>>>>>>
>>> >>>>>>>>>> Jan
>>> >>>>>>>>>>
>>> >>>>>>>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>>> >>>>>>>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <
>>> je.ik@seznam.cz> wrote:
>>> >>>>>>>>>>>> > I don't see batch vs. streaming as part of the model.
>>> One
>>> >>>>>>>>>>>> can have
>>> >>>>>>>>>>>> microbatch, or even a runner that alternates between
>>> different
>>> >>>>>>>>>>>> modes.
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> Although I understand motivation of this statement, this
>>> project
>>> >>>>>>>>>>>> name is
>>> >>>>>>>>>>>> "Apache Beam: An advanced unified programming model". What
>>> does the
>>> >>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the
>>> model?
>>> >>>>>>>>>>> What I mean is that streaming vs. batch is no longer part of
>>> the
>>> >>>>>>>>>>> model
>>> >>>>>>>>>>> (or ideally API), but pushed down to be a concern of the
>>> runner
>>> >>>>>>>>>>> (executor) of the pipeline.
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>
>>> >>>>>>>>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <
>>> je.ik@seznam.cz>
>>> >>>>>>>>>>> wrote:
>>> >>>>>>>>>>>> Hi Kenn,
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> OK, so if we introduce annotation, we can have stateful
>>> ParDo
>>> >>>>>>>>>>>> with sorting, that would perfectly resolve my issues. I
>>> still
>>> >>>>>>>>>>>> have some doubts, though. Let me explain. The current
>>> behavior of
>>> >>>>>>>>>>>> stateful ParDo has the following properties:
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> a) might fail in batch, although runs fine in streaming
>>> (that
>>> >>>>>>>>>>>> is due to the buffering, and unbounded lateness in batch,
>>> which
>>> >>>>>>>>>>>> was discussed back and forth in this thread)
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> b) might be non deterministic (this is because the
>>> elements
>>> >>>>>>>>>>>> arrive at somewhat random order, and even if you do the
>>> operation
>>> >>>>>>>>>>>> "assign unique ID to elements" this might produce different
>>> >>>>>>>>>>>> results when run multiple times)
>>> >>>>>>>>>>> PCollections are *explicitly* unordered. Any operations that
>>> >>>>>>>>>>> assume or
>>> >>>>>>>>>>> depend on a specific ordering for correctness (or
>>> determinism) must
>>> >>>>>>>>>>> provide that ordering themselves (i.e. tolerate "arbitrary
>>> shuffling
>>> >>>>>>>>>>> of inputs"). As you point out, that may be very expensive if
>>> you have
>>> >>>>>>>>>>> very hot keys with very large (unbounded) timestamp skew.
>>> >>>>>>>>>>>
>>> >>>>>>>>>>> StatefulDoFns are low-level operations that should be used
>>> with care;
>>> >>>>>>>>>>> the simpler windowing model gives determinism in the face of
>>> >>>>>>>>>>> unordered
>>> >>>>>>>>>>> data (though late data and non-end-of-window triggering
>>> introduces
>>> >>>>>>>>>>> some of the non-determanism back in).
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>> What worries me most is the property b), because it seems
>>> to me
>>> >>>>>>>>>>>> to have serious consequences - not only that if you run
>>> twice
>>> >>>>>>>>>>>> batch pipeline you would get different results, but even on
>>> >>>>>>>>>>>> streaming, when pipeline fails and gets restarted from
>>> >>>>>>>>>>>> checkpoint, produced output might differ from the previous
>>> run
>>> >>>>>>>>>>>> and data from the first run might have already been
>>> persisted
>>> >>>>>>>>>>>> into sink. That would create somewhat messy outputs.
>>> >>>>>>>>>>> Beam has an exactly-once model. If the data was consumed,
>>> state
>>> >>>>>>>>>>> mutated, and outputs written downstream (these three are
>>> committed
>>> >>>>>>>>>>> together atomically) it will not be replayed. That does not,
>>> of
>>> >>>>>>>>>>> course, solve the non-determanism due to ordering (including
>>> the fact
>>> >>>>>>>>>>> that two operations reading the same PCollection may view
>>> different
>>> >>>>>>>>>>> ordering).
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>> These two properties makes me think that the current
>>> >>>>>>>>>>>> implementation is more of a _special case_ than the general
>>> one.
>>> >>>>>>>>>>>> The general one would be that your state doesn't have the
>>> >>>>>>>>>>>> properties to be able to tolerate buffering problems and/or
>>> >>>>>>>>>>>> non-determinism. Which is the case where you need sorting
>>> in both
>>> >>>>>>>>>>>> streaming and batch to be part of the model.
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> Let me point out one more analogy - that is merging vs.
>>> >>>>>>>>>>>> non-merging windows. The general case (merging windows)
>>> implies
>>> >>>>>>>>>>>> sorting by timestamp in both batch case (explicit) and
>>> streaming
>>> >>>>>>>>>>>> (buffering). The special case (non-merging windows) doesn't
>>> rely
>>> >>>>>>>>>>>> on any timestamp ordering, so the sorting and buffering can
>>> be
>>> >>>>>>>>>>>> dropped. The underlying root cause of this is the same for
>>> both
>>> >>>>>>>>>>>> stateful ParDo and windowing (essentially, assigning window
>>> >>>>>>>>>>>> labels is a stateful operation when windowing function is
>>> merging).
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> The reason for the current behavior of stateful ParDo seems
>>> to be
>>> >>>>>>>>>>>> performance, but is it right to abandon correctness in
>>> favor of
>>> >>>>>>>>>>>> performance? Wouldn't it be more consistent to have the
>>> default
>>> >>>>>>>>>>>> behavior prefer correctness and when you have the specific
>>> >>>>>>>>>>>> conditions of state function having special properties,
>>> then you
>>> >>>>>>>>>>>> can annotate your DoFn (with something like
>>> >>>>>>>>>>>> @TimeOrderingAgnostic), which would yield a better
>>> performance in
>>> >>>>>>>>>>>> that case?
>>> >>>>>>>>>>> There are two separable questions here.
>>> >>>>>>>>>>>
>>> >>>>>>>>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are
>>> received
>>> >>>>>>>>>>> in a specific order? In the current model, it is not. Being
>>> able to
>>> >>>>>>>>>>> read, handle, and produced out-of-order data, including late
>>> data, is
>>> >>>>>>>>>>> a pretty fundamental property of distributed systems.
>>> >>>>>>>>>>>
>>> >>>>>>>>>>> 2) Given that some operations are easier (or possibly only
>>> possible)
>>> >>>>>>>>>>> to write when operating on ordered data, and that different
>>> runners
>>> >>>>>>>>>>> may have (significantly) cheaper ways to provide this
>>> ordering than
>>> >>>>>>>>>>> can be done by the user themselves, should we elevate this
>>> to a
>>> >>>>>>>>>>> property of (Stateful?)DoFns that the runner can provide? I
>>> think a
>>> >>>>>>>>>>> compelling argument can be made here that we should.
>>> >>>>>>>>>>>
>>> >>>>>>>>>>> - Robert
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>
>>> >>>>>>>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> Thanks for the nice small example of a calculation that
>>> depends
>>> >>>>>>>>>>>> on order. You are right that many state machines have this
>>> >>>>>>>>>>>> property. I agree w/ you and Luke that it is convenient for
>>> batch
>>> >>>>>>>>>>>> processing to sort by event timestamp before running a
>>> stateful
>>> >>>>>>>>>>>> ParDo. In streaming you could also implement "sort by event
>>> >>>>>>>>>>>> timestamp" by buffering until you know all earlier data
>>> will be
>>> >>>>>>>>>>>> dropped - a slack buffer up to allowed lateness.
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> I do not think that it is OK to sort in batch and not in
>>> >>>>>>>>>>>> streaming. Many state machines diverge very rapidly when
>>> things
>>> >>>>>>>>>>>> are out of order. So each runner if they see the
>>> >>>>>>>>>>>> "@OrderByTimestamp" annotation (or whatever) needs to
>>> deliver
>>> >>>>>>>>>>>> sorted data (by some mix of buffering and dropping), or to
>>> reject
>>> >>>>>>>>>>>> the pipeline as unsupported.
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> And also want to say that this is not the default case -
>>> many
>>> >>>>>>>>>>>> uses of state & timers in ParDo yield different results at
>>> the
>>> >>>>>>>>>>>> element level, but the results are equivalent at in the big
>>> >>>>>>>>>>>> picture. Such as the example of "assign a unique sequence
>>> number
>>> >>>>>>>>>>>> to each element" or "group into batches" it doesn't matter
>>> >>>>>>>>>>>> exactly what the result is, only that it meets the spec. And
>>> >>>>>>>>>>>> other cases like user funnels are monotonic enough that you
>>> also
>>> >>>>>>>>>>>> don't actually need sorting.
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> Kenn
>>> >>>>>>>>>>>>
>>> >>>>>>>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <
>>> je.ik@seznam.cz>
>>> >>>>>>>>>>>> wrote:
>>> >>>>>>>>>>>>> Yes, the problem will arise probably mostly when you have
>>> not
>>> >>>>>>>>>>>>> well distributed keys (or too few keys). I'm really not
>>> sure if
>>> >>>>>>>>>>>>> a pure GBK with a trigger can solve this - it might help
>>> to have
>>> >>>>>>>>>>>>> data driven trigger. There would still be some doubts,
>>> though.
>>> >>>>>>>>>>>>> The main question is still here - people say, that sorting
>>> by
>>> >>>>>>>>>>>>> timestamp before stateful ParDo would be prohibitively
>>> slow, but
>>> >>>>>>>>>>>>> I don't really see why - the sorting is very probably
>>> already
>>> >>>>>>>>>>>>> there. And if not (hash grouping instead of sorted
>>> grouping),
>>> >>>>>>>>>>>>> then the sorting would affect only user defined
>>> StatefulParDos.
>>> >>>>>>>>>>>>>
>>> >>>>>>>>>>>>> This would suggest that the best way out of this would be
>>> really
>>> >>>>>>>>>>>>> to add annotation, so that the author of the pipeline can
>>> decide.
>>> >>>>>>>>>>>>>
>>> >>>>>>>>>>>>> If that would be acceptable I think I can try to prepare
>>> some
>>> >>>>>>>>>>>>> basic functionality, but I'm not sure, if I would be able
>>> to
>>> >>>>>>>>>>>>> cover all runners / sdks.
>>> >>>>>>>>>>>>>
>>> >>>>>>>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>>> >>>>>>>>>>>>>
>>> >>>>>>>>>>>>> It is read all per key and window and not just read all
>>> (this
>>> >>>>>>>>>>>>> still won't scale with hot keys in the global window). The
>>> GBK
>>> >>>>>>>>>>>>> preceding the StatefulParDo will guarantee that you are
>>> >>>>>>>>>>>>> processing all the values for a specific key and window at
>>> any
>>> >>>>>>>>>>>>> given time. Is there a specific window/trigger that is
>>> missing
>>> >>>>>>>>>>>>> that you feel would remove the need for you to use
>>> StatefulParDo?
>>> >>>>>>>>>>>>>
>>> >>>>>>>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <
>>> je.ik@seznam.cz>
>>> >>>>>>>>>>>>> wrote:
>>> >>>>>>>>>>>>>> Hi Lukasz,
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Today, if you must have a strict order, you must
>>> guarantee
>>> >>>>>>>>>>>>>>> that your StatefulParDo implements the necessary
>>> "buffering &
>>> >>>>>>>>>>>>>>> sorting" into state.
>>> >>>>>>>>>>>>>> Yes, no problem with that. But this whole discussion
>>> started,
>>> >>>>>>>>>>>>>> because *this doesn't work on batch*. You simply cannot
>>> first
>>> >>>>>>>>>>>>>> read everything from distributed storage and then buffer
>>> it all
>>> >>>>>>>>>>>>>> into memory, just to read it again, but sorted. That will
>>> not
>>> >>>>>>>>>>>>>> work. And even if it would, it would be a terrible waste
>>> of
>>> >>>>>>>>>>>>>> resources.
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>> Jan
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <
>>> je.ik@seznam.cz>
>>> >>>>>>>>>>>>>> wrote:
>>> >>>>>>>>>>>>>>> This discussion brings many really interesting questions
>>> for
>>> >>>>>>>>>>>>>>> me. :-)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > I don't see batch vs. streaming as part of the
>>> model. One
>>> >>>>>>>>>>>>>>> can have
>>> >>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
>>> different
>>> >>>>>>>>>>>>>>> modes.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Although I understand motivation of this statement, this
>>> >>>>>>>>>>>>>>> project name is
>>> >>>>>>>>>>>>>>> "Apache Beam: An advanced unified programming model".
>>> What
>>> >>>>>>>>>>>>>>> does the
>>> >>>>>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the
>>> model?
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Using microbatching, chaining of batch jobs, or pure
>>> streaming
>>> >>>>>>>>>>>>>>> are
>>> >>>>>>>>>>>>>>> exactly the "runtime conditions/characteristics" I refer
>>> to.
>>> >>>>>>>>>>>>>>> All these
>>> >>>>>>>>>>>>>>> define several runtime parameters, which in turn define
>>> how
>>> >>>>>>>>>>>>>>> well/badly
>>> >>>>>>>>>>>>>>> will the pipeline perform and how many resources might be
>>> >>>>>>>>>>>>>>> needed. From
>>> >>>>>>>>>>>>>>> my point of view, pure streaming should be the most
>>> resource
>>> >>>>>>>>>>>>>>> demanding
>>> >>>>>>>>>>>>>>> (if not, why bother with batch? why not run everything in
>>> >>>>>>>>>>>>>>> streaming
>>> >>>>>>>>>>>>>>> only? what will there remain to "unify"?).
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > Fortunately, for batch, only the state for a
>>> single key
>>> >>>>>>>>>>>>>>> needs to be
>>> >>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys
>>> across
>>> >>>>>>>>>>>>>>> the range
>>> >>>>>>>>>>>>>>> of skew. Of course if you have few or hot keys, one can
>>> still
>>> >>>>>>>>>>>>>>> have
>>> >>>>>>>>>>>>>>> issues (and this is not specific to StatefulDoFns).
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Yes, but here is still the presumption that my stateful
>>> DoFn can
>>> >>>>>>>>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain
>>> the use
>>> >>>>>>>>>>>>>>> case in
>>> >>>>>>>>>>>>>>> more detail.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Suppose you have input stream consisting of 1s and 0s
>>> (and
>>> >>>>>>>>>>>>>>> some key for
>>> >>>>>>>>>>>>>>> each element, which is irrelevant for the
>>> demonstration). Your
>>> >>>>>>>>>>>>>>> task is
>>> >>>>>>>>>>>>>>> to calculate in running global window the actual number
>>> of
>>> >>>>>>>>>>>>>>> changes
>>> >>>>>>>>>>>>>>> between state 0 and state 1 and vice versa. When the
>>> state
>>> >>>>>>>>>>>>>>> doesn't
>>> >>>>>>>>>>>>>>> change, you don't calculate anything. If input (for
>>> given key)
>>> >>>>>>>>>>>>>>> would be
>>> >>>>>>>>>>>>>>> (tN denotes timestamp N):
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t1: 1
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t2: 0
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t3: 0
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t4: 1
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t5: 1
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t6: 0
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> then the output should yield (supposing that default
>>> state is
>>> >>>>>>>>>>>>>>> zero):
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t1: (one: 1, zero: 0)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t2: (one: 1, zero: 1)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t3: (one: 1, zero: 1)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t4: (one: 2, zero: 1)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t5: (one: 2, zero: 1)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> t6: (one: 2, zero: 2)
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> How would you implement this in current Beam semantics?
>>> >>>>>>>>>>>>>> I think your saying here that I know that my input is
>>> ordered
>>> >>>>>>>>>>>>>> in a specific way and since I assume the order when
>>> writing my
>>> >>>>>>>>>>>>>> pipeline I can perform this optimization. But there is
>>> nothing
>>> >>>>>>>>>>>>>> preventing a runner from noticing that your processing in
>>> the
>>> >>>>>>>>>>>>>> global window with a specific type of trigger and
>>> re-ordering
>>> >>>>>>>>>>>>>> your inputs/processing to get better performance (since
>>> you
>>> >>>>>>>>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>>> >>>>>>>>>>>>>> streaming for the GlobalWindow).
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>> Today, if you must have a strict order, you must
>>> guarantee that
>>> >>>>>>>>>>>>>> your StatefulParDo implements the necessary "buffering &
>>> >>>>>>>>>>>>>> sorting" into state. I can see why you would want an
>>> annotation
>>> >>>>>>>>>>>>>> that says I must have timestamp ordered elements, since it
>>> >>>>>>>>>>>>>> makes writing certain StatefulParDos much easier.
>>> StatefulParDo
>>> >>>>>>>>>>>>>> is a low-level function, it really is the "here you go
>>> and do
>>> >>>>>>>>>>>>>> whatever you need to but here be dragons" function while
>>> >>>>>>>>>>>>>> windowing and triggering is meant to keep many people from
>>> >>>>>>>>>>>>>> writing StatefulParDo in the first place.
>>> >>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > Pipelines that fail in the "worst case" batch
>>> scenario
>>> >>>>>>>>>>>>>>> are likely to
>>> >>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the
>>> watermark
>>> >>>>>>>>>>>>>>> falls
>>> >>>>>>>>>>>>>>> behind in streaming mode as well.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> But the worst case is defined by input of size (available
>>> >>>>>>>>>>>>>>> resources +
>>> >>>>>>>>>>>>>>> single byte) -> pipeline fail. Although it could have
>>> >>>>>>>>>>>>>>> finished, given
>>> >>>>>>>>>>>>>>> the right conditions.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > This might be reasonable, implemented by default by
>>> >>>>>>>>>>>>>>> buffering
>>> >>>>>>>>>>>>>>> everything and releasing elements as the watermark
>>> (+lateness)
>>> >>>>>>>>>>>>>>> advances,
>>> >>>>>>>>>>>>>>> but would likely lead to inefficient (though *maybe*
>>> easier to
>>> >>>>>>>>>>>>>>> reason
>>> >>>>>>>>>>>>>>> about) code.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Sure, the pipeline will be less efficient, because it
>>> would
>>> >>>>>>>>>>>>>>> have to
>>> >>>>>>>>>>>>>>> buffer and sort the inputs. But at least it will produce
>>> >>>>>>>>>>>>>>> correct results
>>> >>>>>>>>>>>>>>> in cases where updates to state are order-sensitive.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > Would it be roughly equivalent to GBK +
>>> FlatMap(lambda
>>> >>>>>>>>>>>>>>> (key, values):
>>> >>>>>>>>>>>>>>> [(key, value) for value in values])?
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> I'd say roughly yes, but difference would be in the
>>> trigger.
>>> >>>>>>>>>>>>>>> The trigger
>>> >>>>>>>>>>>>>>> should ideally fire as soon as watermark (+lateness)
>>> crosses
>>> >>>>>>>>>>>>>>> element
>>> >>>>>>>>>>>>>>> with lowest timestamp in the buffer. Although this could
>>> be
>>> >>>>>>>>>>>>>>> somehow
>>> >>>>>>>>>>>>>>> emulated by fixed trigger each X millis.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> > Or is the underlying desire just to be able to
>>> hint to
>>> >>>>>>>>>>>>>>> the runner
>>> >>>>>>>>>>>>>>> that the code may perform better (e.g. require less
>>> resources)
>>> >>>>>>>>>>>>>>> as skew
>>> >>>>>>>>>>>>>>> is reduced (and hence to order by timestamp iff it's
>>> cheap)?
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> No, the sorting would have to be done in streaming case
>>> as
>>> >>>>>>>>>>>>>>> well. That is
>>> >>>>>>>>>>>>>>> an imperative of the unified model. I think it is
>>> possible to
>>> >>>>>>>>>>>>>>> sort by
>>> >>>>>>>>>>>>>>> timestamp only in batch case (and do it for *all* batch
>>> >>>>>>>>>>>>>>> stateful pardos
>>> >>>>>>>>>>>>>>> without annotation), or introduce annotation, but then
>>> make
>>> >>>>>>>>>>>>>>> the same
>>> >>>>>>>>>>>>>>> guarantees for streaming case as well.
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> Jan
>>> >>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>>> >>>>>>>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>>> >>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>>> >>>>>>>>>>>>>>>>> Hi Robert,
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> yes, I think you rephrased my point - although no
>>> *explicit*
>>> >>>>>>>>>>>>>>>>> guarantees
>>> >>>>>>>>>>>>>>>>> of ordering are given in either mode, there is
>>> *implicit*
>>> >>>>>>>>>>>>>>>>> ordering in
>>> >>>>>>>>>>>>>>>>> streaming case that is due to nature of the processing
>>> - the
>>> >>>>>>>>>>>>>>>>> difference
>>> >>>>>>>>>>>>>>>>> between watermark and timestamp of elements flowing
>>> through
>>> >>>>>>>>>>>>>>>>> the pipeline
>>> >>>>>>>>>>>>>>>>> are generally low (too high difference leads to the
>>> >>>>>>>>>>>>>>>>> overbuffering
>>> >>>>>>>>>>>>>>>>> problem), but there is no such bound on batch.
>>> >>>>>>>>>>>>>>>> Fortunately, for batch, only the state for a single key
>>> needs
>>> >>>>>>>>>>>>>>>> to be
>>> >>>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys
>>> >>>>>>>>>>>>>>>> across the
>>> >>>>>>>>>>>>>>>> range of skew. Of course if you have few or hot keys,
>>> one can
>>> >>>>>>>>>>>>>>>> still
>>> >>>>>>>>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>>> >>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> As a result, I see a few possible solutions:
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> - the best and most natural seems to be
>>> extension of
>>> >>>>>>>>>>>>>>>>> the model, so
>>> >>>>>>>>>>>>>>>>> that it defines batch as not only "streaming pipeline
>>> >>>>>>>>>>>>>>>>> executed in batch
>>> >>>>>>>>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>>> >>>>>>>>>>>>>>>>> characteristics as
>>> >>>>>>>>>>>>>>>>> in streaming case, executed in batch fashion", I really
>>> >>>>>>>>>>>>>>>>> don't think that
>>> >>>>>>>>>>>>>>>>> there are any conflicts with the current model, or
>>> that this
>>> >>>>>>>>>>>>>>>>> could
>>> >>>>>>>>>>>>>>>>> affect performance, because the required sorting (as
>>> pointed by
>>> >>>>>>>>>>>>>>>>> Aljoscha) is very probably already done during
>>> translation
>>> >>>>>>>>>>>>>>>>> of stateful
>>> >>>>>>>>>>>>>>>>> pardos. Also note that this definition only affects
>>> user
>>> >>>>>>>>>>>>>>>>> defined
>>> >>>>>>>>>>>>>>>>> stateful pardos
>>> >>>>>>>>>>>>>>>> I don't see batch vs. streaming as part of the model.
>>> One can
>>> >>>>>>>>>>>>>>>> have
>>> >>>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
>>> >>>>>>>>>>>>>>>> different modes.
>>> >>>>>>>>>>>>>>>> The model describes what the valid outputs are given a
>>> >>>>>>>>>>>>>>>> (sometimes
>>> >>>>>>>>>>>>>>>> partial) set of inputs. It becomes really hard to define
>>> >>>>>>>>>>>>>>>> things like
>>> >>>>>>>>>>>>>>>> "as good runtime characteristics." Once you allow any
>>> >>>>>>>>>>>>>>>> out-of-orderedness, it is not very feasible to try and
>>> define
>>> >>>>>>>>>>>>>>>> (and
>>> >>>>>>>>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>>> >>>>>>>>>>>>>>>> out-of-orderedness.
>>> >>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>> Pipelines that fail in the "worst case" batch scenario
>>> are
>>> >>>>>>>>>>>>>>>> likely to
>>> >>>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the
>>> watermark
>>> >>>>>>>>>>>>>>>> falls
>>> >>>>>>>>>>>>>>>> behind in streaming mode as well.
>>> >>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> - another option would be to introduce
>>> annotation for
>>> >>>>>>>>>>>>>>>>> DoFns (e.g.
>>> >>>>>>>>>>>>>>>>> @RequiresStableTimeCharacteristics), which would
>>> result in
>>> >>>>>>>>>>>>>>>>> the sorting
>>> >>>>>>>>>>>>>>>>> in batch case - but - this extension would have to
>>> ensure
>>> >>>>>>>>>>>>>>>>> the sorting in
>>> >>>>>>>>>>>>>>>>> streaming mode also - it would require definition of
>>> allowed
>>> >>>>>>>>>>>>>>>>> lateness,
>>> >>>>>>>>>>>>>>>>> and triggger (essentially similar to window)
>>> >>>>>>>>>>>>>>>> This might be reasonable, implemented by default by
>>> buffering
>>> >>>>>>>>>>>>>>>> everything and releasing elements as the watermark
>>> (+lateness)
>>> >>>>>>>>>>>>>>>> advances, but would likely lead to inefficient (though
>>> >>>>>>>>>>>>>>>> *maybe* easier
>>> >>>>>>>>>>>>>>>> to reason about) code. Not sure about the semantics of
>>> >>>>>>>>>>>>>>>> triggering
>>> >>>>>>>>>>>>>>>> here, especially data-driven triggers. Would it be
>>> roughly
>>> >>>>>>>>>>>>>>>> equivalent
>>> >>>>>>>>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
>>> >>>>>>>>>>>>>>>> value in
>>> >>>>>>>>>>>>>>>> values])?
>>> >>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>> Or is the underlying desire just to be able to hint to
>>> the
>>> >>>>>>>>>>>>>>>> runner that
>>> >>>>>>>>>>>>>>>> the code may perform better (e.g. require less
>>> resources) as
>>> >>>>>>>>>>>>>>>> skew is
>>> >>>>>>>>>>>>>>>> reduced (and hence to order by timestamp iff it's
>>> cheap)?
>>> >>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> - last option would be to introduce these
>>> "higher order
>>> >>>>>>>>>>>>>>>>> guarantees" in
>>> >>>>>>>>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to
>>> be the
>>> >>>>>>>>>>>>>>>>> worst
>>> >>>>>>>>>>>>>>>>> option to me
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> I see the first two options quite equally good,
>>> although the
>>> >>>>>>>>>>>>>>>>> letter one
>>> >>>>>>>>>>>>>>>>> is probably more time consuming to implement. But it
>>> would
>>> >>>>>>>>>>>>>>>>> bring
>>> >>>>>>>>>>>>>>>>> additional feature to streaming case as well.
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> Thanks for any thoughts.
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> Jan
>>> >>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>>> >>>>>>>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>>> >>>>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>>> >>>>>>>>>>>>>>>>>>> Hi Reuven,
>>> >>>>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
>>> >>>>>>>>>>>>>>>>>>>> runners.
>>> >>>>>>>>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic
>>> inside
>>> >>>>>>>>>>>>>>>>>>> the state works for absolutely unbounded
>>> out-of-orderness
>>> >>>>>>>>>>>>>>>>>>> of elements. That basically (practically) can work
>>> only
>>> >>>>>>>>>>>>>>>>>>> for cases, where the order of input elements doesn't
>>> >>>>>>>>>>>>>>>>>>> matter. But, "state" can refer to "state machine",
>>> and any
>>> >>>>>>>>>>>>>>>>>>> time you have a state machine involved, then the
>>> ordering
>>> >>>>>>>>>>>>>>>>>>> of elements would matter.
>>> >>>>>>>>>>>>>>>>>> No guarantees on order are provided in *either*
>>> streaming
>>> >>>>>>>>>>>>>>>>>> or batch
>>> >>>>>>>>>>>>>>>>>> mode by the model. However, it is the case that in
>>> order to
>>> >>>>>>>>>>>>>>>>>> make
>>> >>>>>>>>>>>>>>>>>> forward progress most streaming runners attempt to
>>> limit
>>> >>>>>>>>>>>>>>>>>> the amount of
>>> >>>>>>>>>>>>>>>>>> out-of-orderedness of elements (in terms of event
>>> time vs.
>>> >>>>>>>>>>>>>>>>>> processing
>>> >>>>>>>>>>>>>>>>>> time) to make forward progress, which in turn could
>>> help
>>> >>>>>>>>>>>>>>>>>> cap the
>>> >>>>>>>>>>>>>>>>>> amount of state that must be held concurrently,
>>> whereas a
>>> >>>>>>>>>>>>>>>>>> batch runner
>>> >>>>>>>>>>>>>>>>>> may not allow any state to be safely discarded until
>>> the whole
>>> >>>>>>>>>>>>>>>>>> timeline from infinite past to infinite future has
>>> been
>>> >>>>>>>>>>>>>>>>>> observed.
>>> >>>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
>>> >>>>>>>>>>>>>>>>>> batch" in batch mode.
>>> >>>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>>> >>>>>>>>>>>>>>>>>> <mx...@apache.org> wrote:
>>> >>>>>>>>>>>>>>>>>>
>>> >>>>>>>>>>>>>>>>>>>> batch semantics and streaming semantics
>>> differs only
>>> >>>>>>>>>>>>>>>>>>>> in that I can have GlobalWindow with default
>>> trigger on
>>> >>>>>>>>>>>>>>>>>>>> batch and cannot on stream
>>> >>>>>>>>>>>>>>>>>>> You can have a GlobalWindow in streaming with a
>>> default
>>> >>>>>>>>>>>>>>>>>>> trigger. You
>>> >>>>>>>>>>>>>>>>>>> could define additional triggers that do early
>>> firings.
>>> >>>>>>>>>>>>>>>>>>> And you could
>>> >>>>>>>>>>>>>>>>>>> even trigger the global window by advancing the
>>> watermark
>>> >>>>>>>>>>>>>>>>>>> to +inf.
>>> >>>>>>>>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global
>>> window with
>>> >>>>>>>>>>>>>>>>>> default
>>> >>>>>>>>>>>>>>>>>> trigger on unbounded PCollections in the SDK because
>>> this
>>> >>>>>>>>>>>>>>>>>> is more
>>> >>>>>>>>>>>>>>>>>> likely to be user error than an actual desire to have
>>> no
>>> >>>>>>>>>>>>>>>>>> output until
>>> >>>>>>>>>>>>>>>>>> drain. But it's semantically valid in the model.
>>>
>>>
Re: Definition of Unified model
Posted by Jan Lukavský <je...@seznam.cz>.
Could this be solved by "expanding the dimensionality" of time field?
What I mean by that - if input element to to FlatMap has sequence number
T, then the (stateless) FlatMap knows the ordering of output elements,
right? If it would expand the field by which it will next sort the
elements to (X, 1), (X, 2), ... (X, N), then it would be possible to
sort the elements back later. There seems to be no need for state to
achieve that, or?
Jan
On 5/28/19 6:52 PM, Reuven Lax wrote:
> A slightly larger concern: it also will force users to create stateful
> DoFns everywhere to generate these sequence numbers. If I have a ParDo
> that is not a simple 1:1 transform (i.e. not MapElements), then the
> ParDo will need to generate its own sequence numbers for ordering, and
> the only safe way to do so is to use a stateful DoFn. This turns what
> used to be a simple in-memory DoFn into one that has to access state.
> Also I believe many runners will not fuse stateful DoFns. While none
> of this poses a problem for the model, it could make ordering
> extremely expensive to achieve.
>
> Reuven
>
> On Tue, May 28, 2019 at 6:09 AM Jan Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
>
> Hi Reuven,
>
> > It also gets awkward with Flatten - the sequence number is no
> longer enough, you must also encode which side of the flatten each
> element came from.
>
> That is a generic need. Even if you read data from Kafka, the
> offsets are comparable only inside single partition. So, for Kafka
> to work as a FIFO for ordering, elements with same key have to be
> pushed to the same partition (otherwise Kafka cannot act as FIFO,
> because different partitions can be handled by different brokers,
> which means different observers and they therefore might not agree
> on the order of events). So if we want to emulate FIFO per key,
> then the sequence IDs have also be per key.
>
> On 5/28/19 2:33 PM, Reuven Lax wrote:
>> Sequence metadata does have the disadvantage that users can no
>> longer use the types coming from the source. You must create a
>> new type that contains a sequence number (unless Beam provides
>> this). It also gets awkward with Flatten - the sequence number is
>> no longer enough, you must also encode which side of the flatten
>> each element came from.
>>
>> On Tue, May 28, 2019 at 3:18 AM Jan Lukavský <je.ik@seznam.cz
>> <ma...@seznam.cz>> wrote:
>>
>> As I understood it, Kenn was supporting the idea that
>> sequence metadata
>> is preferable over FIFO. I was trying to point out, that it
>> even should
>> provide the same functionally as FIFO, plus one important more -
>> reproducibility and ability to being persisted and reused the
>> same way
>> in batch and streaming.
>>
>> There is no doubt, that sequence metadata can be stored in every
>> storage. But, regarding some implicit ordering that sources
>> might have -
>> yes, of course, data written into HDFS or Cloud Storage has
>> ordering,
>> but only partial - inside some bulk (e.g. file) and the
>> ordering is not
>> defined correctly on boundaries of these bulks (between
>> files). That is
>> why I'd say, that ordering of sources is relevant only for
>> (partitioned!) streaming sources and generally always reduces to
>> sequence metadata (e.g. offsets).
>>
>> Jan
>>
>> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
>> > Huge +1 to all Kenn said.
>> >
>> > Jan, batch sources can have orderings too, just like Kafka.
>> I think
>> > it's reasonable (for both batch and streaming) that if a
>> source has an
>> > ordering that is an important part of the data, it should
>> preserve
>> > this ordering into the data itself (e.g. as sequence
>> numbers, offsets,
>> > etc.)
>> >
>> > On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles
>> <kenn@apache.org <ma...@apache.org>> wrote:
>> >> I strongly prefer explicit sequence metadata over FIFO
>> requirements, because:
>> >>
>> >> - FIFO is complex to specify: for example Dataflow has
>> "per stage key-to-key" FIFO today, but it is not guaranteed
>> to remain so (plus "stage" is not a portable concept, nor
>> even guaranteed to remain a Dataflow concept)
>> >> - complex specifications are by definition poor
>> usability (if necessary, then it is what it is)
>> >> - overly restricts the runner, reduces parallelism, for
>> example any non-stateful ParDo has per-element parallelism,
>> not per "key"
>> >> - another perspective on that: FIFO makes everyone pay
>> rather than just the transform that requires exactly sequencing
>> >> - previous implementation details like reshuffles become
>> part of the model
>> >> - I'm not even convinced the use cases involved are
>> addressed by some careful FIFO restrictions; many sinks
>> re-key and they would all have to become aware of how keying
>> of a sequence of "stages" affects the end-to-end FIFO
>> >>
>> >> A noop becoming a non-noop is essentially the mathematical
>> definition of moving from higher-level to lower-level
>> abstraction.
>> >>
>> >> So this strikes at the core question of what level of
>> abstraction Beam aims to represent. Lower-level means there
>> are fewer possible implementations and it is more tied to the
>> underlying architecture, and anything not near-exact match
>> pays a huge penalty. Higher-level means there are more
>> implementations possible with different tradeoffs, though
>> they may all pay a minor penalty.
>> >>
>> >> I could be convinced to change my mind, but it needs some
>> extensive design, examples, etc. I think it is probably about
>> the most consequential design decision in the whole Beam
>> model, around the same level as the decision to use ParDo and
>> GBK as the primitives IMO.
>> >>
>> >> Kenn
>> >>
>> >> On Thu, May 23, 2019 at 10:17 AM Reuven Lax
>> <relax@google.com <ma...@google.com>> wrote:
>> >>> Not really. I'm suggesting that some variant of FIFO
>> ordering is necessary, which requires either runners natively
>> support FIFO ordering or transforms adding some extra
>> sequence number to each record to sort by.
>> >>>
>> >>> I still think your proposal is very useful by the way.
>> I'm merely pointing out that to solve the state-machine
>> problem we probably need something more.
>> >>>
>> >>> Reuven
>> >>>
>> >>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský
>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>> >>>> Hi,
>> >>>> yes. It seems that ordering by user supplied UDF makes
>> sense and I will update the design proposal accordingly.
>> >>>> Would that solve the issues you mention?
>> >>>> Jan
>> >>>> ---------- Původní e-mail ----------
>> >>>> Od: Reuven Lax <relax@google.com <ma...@google.com>>
>> >>>> Komu: dev <dev@beam.apache.org <ma...@beam.apache.org>>
>> >>>> Datum: 23. 5. 2019 18:44:38
>> >>>> Předmět: Re: Definition of Unified model
>> >>>>
>> >>>> I'm simply saying that timestamp ordering is
>> insufficient for state machines. I wasn't proposing Kafka as
>> a solution - that was simply an example of how people solve
>> this problem in other scenarios.
>> >>>>
>> >>>> BTW another example of ordering: Imagine today that you
>> have a triggered Sum aggregation writing out to a key-value
>> sink. In theory we provide no ordering, so the sink might
>> write the triggered sums in the wrong order, ending up with
>> an incorrect value in the sink. In this case you probably
>> want values ordered by trigger pane index.
>> >>>>
>> >>>> Reuven
>> >>>>
>> >>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský
>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>> >>>>
>> >>>> Hi Reuven,
>> >>>> I share the view point of Robert. I think the isuue you
>> refer to is not in reality related to timestamps, but to the
>> fact, that ordering of events in time is observer dependent
>> (either caused by relativity, or time skew, essentially this
>> has the same consequences). And the resolution in fact isn't
>> Kafka, but generally an authoritative observer, that tells
>> you "I saw the events in this order". And you either have one
>> (and have the outcome of his observation persisted in the
>> data - e.g. as offset in Kafka partition), then you should be
>> able to use it (maybe that suggests afterall that sorting by
>> some user supplied UDF might make sense), or do not have it,
>> and then any interpretation of the data seems to be equally
>> valid. Although determinism is fine, of course.
>> >>>> Jan
>> >>>> ---------- Původní e-mail ----------
>> >>>> Od: Reuven Lax <relax@google.com <ma...@google.com>>
>> >>>> Komu: dev <dev@beam.apache.org <ma...@beam.apache.org>>
>> >>>> Datum: 23. 5. 2019 17:39:12
>> >>>> Předmět: Re: Definition of Unified model
>> >>>>
>> >>>> So an example would be elements of type
>> "startUserSession" and "endUserSession" (website sessions,
>> not Beam sessions). Logically you may need to process them in
>> the correct order if you have any sort of state-machine
>> logic. However timestamp ordering is never guaranteed to
>> match the logical ordering. Not only might you have several
>> elements with the same timestamp, but in reality time skew
>> across backend servers can cause the events to have
>> timestamps in reverse order of the actual causality order.
>> >>>>
>> >>>> People do solve this problem today though. Publish the
>> events to Kafka, making sure that events for the same user
>> end up in the same Kafka partition. This ensures that the
>> events appear in the Kafka partitions in causality order,
>> even if the timestamp order doesn't match. The your Kafka
>> subscriber simply process the elements in each partition in
>> order.
>> >>>>
>> >>>> I think the ability to impose FIFO causality ordering is
>> what's needed for any state-machine work. Timestamp ordering
>> has advantages (though often I think the advantage is in
>> state), but does not solve this problem.
>> >>>>
>> >>>> Reuven
>> >>>>
>> >>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw
>> <robertwb@google.com <ma...@google.com>> wrote:
>> >>>>
>> >>>> Good point.
>> >>>>
>> >>>> The "implementation-specific" way I would do this is
>> >>>> window-by-instant, followed by a DoFn that gets all the
>> elements with
>> >>>> the same timestamp and sorts/acts accordingly, but this
>> counts on the
>> >>>> runner producing windows in timestamp order (likely?)
>> and also the
>> >>>> subsequent DoFn getting them in this order (also likely,
>> due to
>> >>>> fusion).
>> >>>>
>> >>>> One could make the argument that, though it does not provide
>> >>>> deterministic behavior, getting elements of the same
>> timestamp in
>> >>>> different orders should produce equally valid
>> interpretations of the
>> >>>> data. (After all, due to relatively, timestamps are not
>> technically
>> >>>> well ordered across space.) I can see how data-dependent
>> tiebreakers
>> >>>> could be useful, or promises of preservation of order
>> between
>> >>>> operations.
>> >>>>
>> >>>> - Robert
>> >>>>
>> >>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax
>> <relax@google.com <ma...@google.com>> wrote:
>> >>>>> So Jan's example of state machines is quite a valid use
>> case for ordering. However in my experience, timestamp
>> ordering is insufficient for state machines. Elements that
>> cause state transitions might come in with the exact same
>> timestamp, yet still have a necessary ordering. Especially
>> given Beam's decision to have milliseconds timestamps this is
>> possible, but even at microsecond or nanosecond precision
>> this can happen at scale. To handle state machines you
>> usually need some sort of FIFO ordering along with an ordered
>> sources, such as Kafka, not timestamp ordering.
>> >>>>>
>> >>>>> Reuven
>> >>>>>
>> >>>>> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský
>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>> >>>>>> Hi all,
>> >>>>>>
>> >>>>>> thanks everyone for this discussion. I think I have
>> gathered enough
>> >>>>>> feedback to be able to put down a proposition for
>> changes, which I will
>> >>>>>> do and send to this list for further discussion. There
>> are still doubts
>> >>>>>> remaining the non-determinism and it's relation to
>> outputs stability vs.
>> >>>>>> latency. But I will try to clarify all this in the
>> design document.
>> >>>>>>
>> >>>>>> Thanks,
>> >>>>>>
>> >>>>>> Jan
>> >>>>>>
>> >>>>>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>> >>>>>>>> Someone from Flink might correct me if I'm wrong,
>> but that's my
>> >>>>>>>> current understanding.
>> >>>>>>> In essence your description of how exactly-once works
>> in Flink is
>> >>>>>>> correct. The general assumption in Flink is that
>> pipelines must be
>> >>>>>>> deterministic and thus produce idempotent writes in
>> the case of
>> >>>>>>> failures. However, that doesn't mean Beam sinks can't
>> guarantee a bit
>> >>>>>>> more with what Flink has to offer.
>> >>>>>>>
>> >>>>>>> Luke already mentioned the design discussions for
>> @RequiresStableInput
>> >>>>>>> which ensures idempotent writes for non-deterministic
>> pipelines. This
>> >>>>>>> is not part of the model but an optional Beam feature.
>> >>>>>>>
>> >>>>>>> We recently implemented support for
>> @RequiresStableInput in the Flink
>> >>>>>>> Runner. Reuven mentioned the Flink checkpoint
>> confirmation, which
>> >>>>>>> allows us to buffer (and checkpoint) processed data
>> and only emit it
>> >>>>>>> once a Flink checkpoint has completed.
>> >>>>>>>
>> >>>>>>> Cheers,
>> >>>>>>> Max
>> >>>>>>>
>> >>>>>>> On 21.05.19 16:49, Jan Lukavský wrote:
>> >>>>>>>> Hi,
>> >>>>>>>>
>> >>>>>>>> > Actually, I think it is a larger (open) question
>> whether exactly
>> >>>>>>>> once is guaranteed by the model or whether runners
>> are allowed to
>> >>>>>>>> relax that. I would think, however, that sources
>> correctly
>> >>>>>>>> implemented should be idempotent when run atop an
>> exactly once
>> >>>>>>>> infrastructure such as Flink of Dataflow.
>> >>>>>>>>
>> >>>>>>>> I would assume, that the model basically inherits
>> guarantees of
>> >>>>>>>> underlying infrastructure. Because Flink does not
>> work as you
>> >>>>>>>> described (atomic commit of inputs, state and
>> outputs), but rather a
>> >>>>>>>> checkpoint mark is flowing through the DAG much like
>> watermark and on
>> >>>>>>>> failures operators are restored and data
>> reprocessed, it (IMHO)
>> >>>>>>>> implies, that you have exactly once everywhere in
>> the DAG *but*
>> >>>>>>>> sinks. That is because sinks cannot be restored to
>> previous state,
>> >>>>>>>> instead sinks are supposed to be idempotent in order
>> for the exactly
>> >>>>>>>> once to really work (or at least be able to commit
>> outputs on
>> >>>>>>>> checkpoint in sink). That implies that if you don't
>> have sink that is
>> >>>>>>>> able to commit outputs atomically on checkpoint, the
>> pipeline
>> >>>>>>>> execution should be deterministic upon retries,
>> otherwise shadow
>> >>>>>>>> writes from failed paths of the pipeline might appear.
>> >>>>>>>>
>> >>>>>>>> Someone from Flink might correct me if I'm wrong,
>> but that's my
>> >>>>>>>> current understanding.
>> >>>>>>>>
>> >>>>>>>> > Sounds like we should make this clearer.
>> >>>>>>>>
>> >>>>>>>> I meant that you are right that we must not in any
>> thoughts we are
>> >>>>>>>> having forget that streams are by definition
>> out-of-order. That is
>> >>>>>>>> property that we cannot change. But - that doesn't
>> limit us from
>> >>>>>>>> creating operator that presents the data to UDF as
>> if the stream was
>> >>>>>>>> ideally sorted. It can do that by introducing
>> latency, of course.
>> >>>>>>>>
>> >>>>>>>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>> >>>>>>>>> Reza: One could provide something like this as a
>> utility class, but
>> >>>>>>>>> one downside is that it is not scale invariant. It
>> requires a tuning
>> >>>>>>>>> parameter that, if to small, won't mitigate the
>> problem, but if to
>> >>>>>>>>> big, greatly increases latency. (Possibly one could
>> define a dynamic
>> >>>>>>>>> session-like window to solve this though...) It
>> also might be harder
>> >>>>>>>>> for runners that *can* cheaply present stuff in
>> timestamp order to
>> >>>>>>>>> optimize. (That and, in practice, our
>> annotation-style process methods
>> >>>>>>>>> don't lend themselves to easy composition.) I think
>> it could work in
>> >>>>>>>>> specific cases though.
>> >>>>>>>>>
>> >>>>>>>>> More inline below.
>> >>>>>>>>>
>> >>>>>>>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský
>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>> >>>>>>>>>> Hi Robert,
>> >>>>>>>>>>
>> >>>>>>>>>> > Beam has an exactly-once model. If the data
>> was consumed, state
>> >>>>>>>>>> mutated, and outputs written downstream (these
>> three are committed
>> >>>>>>>>>> together atomically) it will not be replayed. That
>> does not, of
>> >>>>>>>>>> course,
>> >>>>>>>>>> solve the non-determanism due to ordering
>> (including the fact that two
>> >>>>>>>>>> operations reading the same PCollection may view
>> different ordering).
>> >>>>>>>>>>
>> >>>>>>>>>> I think what you describe is a property of a
>> runner, not of the model,
>> >>>>>>>>>> right? I think if I run my pipeline on Flink I
>> will not get this
>> >>>>>>>>>> atomicity, because although Flink uses also
>> exactly-once model if
>> >>>>>>>>>> might
>> >>>>>>>>>> write outputs multiple times.
>> >>>>>>>>> Actually, I think it is a larger (open) question
>> whether exactly once
>> >>>>>>>>> is guaranteed by the model or whether runners are
>> allowed to relax
>> >>>>>>>>> that. I would think, however, that sources
>> correctly implemented
>> >>>>>>>>> should be idempotent when run atop an exactly once
>> infrastructure such
>> >>>>>>>>> as Flink of Dataflow.
>> >>>>>>>>>
>> >>>>>>>>>> > 1) Is it correct for a (Stateful)DoFn to
>> assume elements are
>> >>>>>>>>>> received
>> >>>>>>>>>> in a specific order? In the current model, it is
>> not. Being able to
>> >>>>>>>>>> read, handle, and produced out-of-order data,
>> including late data,
>> >>>>>>>>>> is a
>> >>>>>>>>>> pretty fundamental property of distributed systems.
>> >>>>>>>>>>
>> >>>>>>>>>> Yes, absolutely. The argument here is not that
>> Stateful ParDo should
>> >>>>>>>>>> presume to receive elements in any order, but to
>> _present_ it as
>> >>>>>>>>>> such to
>> >>>>>>>>>> the user @ProcessElement function.
>> >>>>>>>>> Sounds like we should make this clearer.
>> >>>>>>>>>
>> >>>>>>>>>> > 2) Given that some operations are easier (or
>> possibly only
>> >>>>>>>>>> possible)
>> >>>>>>>>>> to write when operating on ordered data, and that
>> different runners
>> >>>>>>>>>> may
>> >>>>>>>>>> have (significantly) cheaper ways to provide this
>> ordering than can be
>> >>>>>>>>>> done by the user themselves, should we elevate
>> this to a property of
>> >>>>>>>>>> (Stateful?)DoFns that the runner can provide? I
>> think a compelling
>> >>>>>>>>>> argument can be made here that we should.
>> >>>>>>>>>>
>> >>>>>>>>>> +1
>> >>>>>>>>>>
>> >>>>>>>>>> Jan
>> >>>>>>>>>>
>> >>>>>>>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>> >>>>>>>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský
>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
>> >>>>>>>>>>>> > I don't see batch vs. streaming as part of
>> the model. One
>> >>>>>>>>>>>> can have
>> >>>>>>>>>>>> microbatch, or even a runner that alternates
>> between different
>> >>>>>>>>>>>> modes.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Although I understand motivation of this
>> statement, this project
>> >>>>>>>>>>>> name is
>> >>>>>>>>>>>> "Apache Beam: An advanced unified programming
>> model". What does the
>> >>>>>>>>>>>> model unify, if "streaming vs. batch" is not
>> part of the model?
>> >>>>>>>>>>> What I mean is that streaming vs. batch is no
>> longer part of the
>> >>>>>>>>>>> model
>> >>>>>>>>>>> (or ideally API), but pushed down to be a concern
>> of the runner
>> >>>>>>>>>>> (executor) of the pipeline.
>> >>>>>>>>>>>
>> >>>>>>>>>>>
>> >>>>>>>>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský
>> <je.ik@seznam.cz <ma...@seznam.cz>>
>> >>>>>>>>>>> wrote:
>> >>>>>>>>>>>> Hi Kenn,
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> OK, so if we introduce annotation, we can have
>> stateful ParDo
>> >>>>>>>>>>>> with sorting, that would perfectly resolve my
>> issues. I still
>> >>>>>>>>>>>> have some doubts, though. Let me explain. The
>> current behavior of
>> >>>>>>>>>>>> stateful ParDo has the following properties:
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> a) might fail in batch, although runs fine in
>> streaming (that
>> >>>>>>>>>>>> is due to the buffering, and unbounded lateness
>> in batch, which
>> >>>>>>>>>>>> was discussed back and forth in this thread)
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> b) might be non deterministic (this is because
>> the elements
>> >>>>>>>>>>>> arrive at somewhat random order, and even if you
>> do the operation
>> >>>>>>>>>>>> "assign unique ID to elements" this might
>> produce different
>> >>>>>>>>>>>> results when run multiple times)
>> >>>>>>>>>>> PCollections are *explicitly* unordered. Any
>> operations that
>> >>>>>>>>>>> assume or
>> >>>>>>>>>>> depend on a specific ordering for correctness (or
>> determinism) must
>> >>>>>>>>>>> provide that ordering themselves (i.e. tolerate
>> "arbitrary shuffling
>> >>>>>>>>>>> of inputs"). As you point out, that may be very
>> expensive if you have
>> >>>>>>>>>>> very hot keys with very large (unbounded)
>> timestamp skew.
>> >>>>>>>>>>>
>> >>>>>>>>>>> StatefulDoFns are low-level operations that
>> should be used with care;
>> >>>>>>>>>>> the simpler windowing model gives determinism in
>> the face of
>> >>>>>>>>>>> unordered
>> >>>>>>>>>>> data (though late data and non-end-of-window
>> triggering introduces
>> >>>>>>>>>>> some of the non-determanism back in).
>> >>>>>>>>>>>
>> >>>>>>>>>>>> What worries me most is the property b), because
>> it seems to me
>> >>>>>>>>>>>> to have serious consequences - not only that if
>> you run twice
>> >>>>>>>>>>>> batch pipeline you would get different results,
>> but even on
>> >>>>>>>>>>>> streaming, when pipeline fails and gets
>> restarted from
>> >>>>>>>>>>>> checkpoint, produced output might differ from
>> the previous run
>> >>>>>>>>>>>> and data from the first run might have already
>> been persisted
>> >>>>>>>>>>>> into sink. That would create somewhat messy outputs.
>> >>>>>>>>>>> Beam has an exactly-once model. If the data was
>> consumed, state
>> >>>>>>>>>>> mutated, and outputs written downstream (these
>> three are committed
>> >>>>>>>>>>> together atomically) it will not be replayed.
>> That does not, of
>> >>>>>>>>>>> course, solve the non-determanism due to ordering
>> (including the fact
>> >>>>>>>>>>> that two operations reading the same PCollection
>> may view different
>> >>>>>>>>>>> ordering).
>> >>>>>>>>>>>
>> >>>>>>>>>>>> These two properties makes me think that the current
>> >>>>>>>>>>>> implementation is more of a _special case_ than
>> the general one.
>> >>>>>>>>>>>> The general one would be that your state doesn't
>> have the
>> >>>>>>>>>>>> properties to be able to tolerate buffering
>> problems and/or
>> >>>>>>>>>>>> non-determinism. Which is the case where you
>> need sorting in both
>> >>>>>>>>>>>> streaming and batch to be part of the model.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Let me point out one more analogy - that is
>> merging vs.
>> >>>>>>>>>>>> non-merging windows. The general case (merging
>> windows) implies
>> >>>>>>>>>>>> sorting by timestamp in both batch case
>> (explicit) and streaming
>> >>>>>>>>>>>> (buffering). The special case (non-merging
>> windows) doesn't rely
>> >>>>>>>>>>>> on any timestamp ordering, so the sorting and
>> buffering can be
>> >>>>>>>>>>>> dropped. The underlying root cause of this is
>> the same for both
>> >>>>>>>>>>>> stateful ParDo and windowing (essentially,
>> assigning window
>> >>>>>>>>>>>> labels is a stateful operation when windowing
>> function is merging).
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> The reason for the current behavior of stateful
>> ParDo seems to be
>> >>>>>>>>>>>> performance, but is it right to abandon
>> correctness in favor of
>> >>>>>>>>>>>> performance? Wouldn't it be more consistent to
>> have the default
>> >>>>>>>>>>>> behavior prefer correctness and when you have
>> the specific
>> >>>>>>>>>>>> conditions of state function having special
>> properties, then you
>> >>>>>>>>>>>> can annotate your DoFn (with something like
>> >>>>>>>>>>>> @TimeOrderingAgnostic), which would yield a
>> better performance in
>> >>>>>>>>>>>> that case?
>> >>>>>>>>>>> There are two separable questions here.
>> >>>>>>>>>>>
>> >>>>>>>>>>> 1) Is it correct for a (Stateful)DoFn to assume
>> elements are received
>> >>>>>>>>>>> in a specific order? In the current model, it is
>> not. Being able to
>> >>>>>>>>>>> read, handle, and produced out-of-order data,
>> including late data, is
>> >>>>>>>>>>> a pretty fundamental property of distributed systems.
>> >>>>>>>>>>>
>> >>>>>>>>>>> 2) Given that some operations are easier (or
>> possibly only possible)
>> >>>>>>>>>>> to write when operating on ordered data, and that
>> different runners
>> >>>>>>>>>>> may have (significantly) cheaper ways to provide
>> this ordering than
>> >>>>>>>>>>> can be done by the user themselves, should we
>> elevate this to a
>> >>>>>>>>>>> property of (Stateful?)DoFns that the runner can
>> provide? I think a
>> >>>>>>>>>>> compelling argument can be made here that we should.
>> >>>>>>>>>>>
>> >>>>>>>>>>> - Robert
>> >>>>>>>>>>>
>> >>>>>>>>>>>
>> >>>>>>>>>>>
>> >>>>>>>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Thanks for the nice small example of a
>> calculation that depends
>> >>>>>>>>>>>> on order. You are right that many state machines
>> have this
>> >>>>>>>>>>>> property. I agree w/ you and Luke that it is
>> convenient for batch
>> >>>>>>>>>>>> processing to sort by event timestamp before
>> running a stateful
>> >>>>>>>>>>>> ParDo. In streaming you could also implement
>> "sort by event
>> >>>>>>>>>>>> timestamp" by buffering until you know all
>> earlier data will be
>> >>>>>>>>>>>> dropped - a slack buffer up to allowed lateness.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> I do not think that it is OK to sort in batch
>> and not in
>> >>>>>>>>>>>> streaming. Many state machines diverge very
>> rapidly when things
>> >>>>>>>>>>>> are out of order. So each runner if they see the
>> >>>>>>>>>>>> "@OrderByTimestamp" annotation (or whatever)
>> needs to deliver
>> >>>>>>>>>>>> sorted data (by some mix of buffering and
>> dropping), or to reject
>> >>>>>>>>>>>> the pipeline as unsupported.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> And also want to say that this is not the
>> default case - many
>> >>>>>>>>>>>> uses of state & timers in ParDo yield different
>> results at the
>> >>>>>>>>>>>> element level, but the results are equivalent at
>> in the big
>> >>>>>>>>>>>> picture. Such as the example of "assign a unique
>> sequence number
>> >>>>>>>>>>>> to each element" or "group into batches" it
>> doesn't matter
>> >>>>>>>>>>>> exactly what the result is, only that it meets
>> the spec. And
>> >>>>>>>>>>>> other cases like user funnels are monotonic
>> enough that you also
>> >>>>>>>>>>>> don't actually need sorting.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Kenn
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský
>> <je.ik@seznam.cz <ma...@seznam.cz>>
>> >>>>>>>>>>>> wrote:
>> >>>>>>>>>>>>> Yes, the problem will arise probably mostly
>> when you have not
>> >>>>>>>>>>>>> well distributed keys (or too few keys). I'm
>> really not sure if
>> >>>>>>>>>>>>> a pure GBK with a trigger can solve this - it
>> might help to have
>> >>>>>>>>>>>>> data driven trigger. There would still be some
>> doubts, though.
>> >>>>>>>>>>>>> The main question is still here - people say,
>> that sorting by
>> >>>>>>>>>>>>> timestamp before stateful ParDo would be
>> prohibitively slow, but
>> >>>>>>>>>>>>> I don't really see why - the sorting is very
>> probably already
>> >>>>>>>>>>>>> there. And if not (hash grouping instead of
>> sorted grouping),
>> >>>>>>>>>>>>> then the sorting would affect only user defined
>> StatefulParDos.
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> This would suggest that the best way out of
>> this would be really
>> >>>>>>>>>>>>> to add annotation, so that the author of the
>> pipeline can decide.
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> If that would be acceptable I think I can try
>> to prepare some
>> >>>>>>>>>>>>> basic functionality, but I'm not sure, if I
>> would be able to
>> >>>>>>>>>>>>> cover all runners / sdks.
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> It is read all per key and window and not just
>> read all (this
>> >>>>>>>>>>>>> still won't scale with hot keys in the global
>> window). The GBK
>> >>>>>>>>>>>>> preceding the StatefulParDo will guarantee that
>> you are
>> >>>>>>>>>>>>> processing all the values for a specific key
>> and window at any
>> >>>>>>>>>>>>> given time. Is there a specific window/trigger
>> that is missing
>> >>>>>>>>>>>>> that you feel would remove the need for you to
>> use StatefulParDo?
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský
>> <je.ik@seznam.cz <ma...@seznam.cz>>
>> >>>>>>>>>>>>> wrote:
>> >>>>>>>>>>>>>> Hi Lukasz,
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Today, if you must have a strict order, you
>> must guarantee
>> >>>>>>>>>>>>>>> that your StatefulParDo implements the
>> necessary "buffering &
>> >>>>>>>>>>>>>>> sorting" into state.
>> >>>>>>>>>>>>>> Yes, no problem with that. But this whole
>> discussion started,
>> >>>>>>>>>>>>>> because *this doesn't work on batch*. You
>> simply cannot first
>> >>>>>>>>>>>>>> read everything from distributed storage and
>> then buffer it all
>> >>>>>>>>>>>>>> into memory, just to read it again, but
>> sorted. That will not
>> >>>>>>>>>>>>>> work. And even if it would, it would be a
>> terrible waste of
>> >>>>>>>>>>>>>> resources.
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> Jan
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský
>> <je.ik@seznam.cz <ma...@seznam.cz>>
>> >>>>>>>>>>>>>> wrote:
>> >>>>>>>>>>>>>>> This discussion brings many really
>> interesting questions for
>> >>>>>>>>>>>>>>> me. :-)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > I don't see batch vs. streaming as part
>> of the model. One
>> >>>>>>>>>>>>>>> can have
>> >>>>>>>>>>>>>>> microbatch, or even a runner that alternates
>> between different
>> >>>>>>>>>>>>>>> modes.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Although I understand motivation of this
>> statement, this
>> >>>>>>>>>>>>>>> project name is
>> >>>>>>>>>>>>>>> "Apache Beam: An advanced unified programming
>> model". What
>> >>>>>>>>>>>>>>> does the
>> >>>>>>>>>>>>>>> model unify, if "streaming vs. batch" is not
>> part of the model?
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Using microbatching, chaining of batch jobs,
>> or pure streaming
>> >>>>>>>>>>>>>>> are
>> >>>>>>>>>>>>>>> exactly the "runtime
>> conditions/characteristics" I refer to.
>> >>>>>>>>>>>>>>> All these
>> >>>>>>>>>>>>>>> define several runtime parameters, which in
>> turn define how
>> >>>>>>>>>>>>>>> well/badly
>> >>>>>>>>>>>>>>> will the pipeline perform and how many
>> resources might be
>> >>>>>>>>>>>>>>> needed. From
>> >>>>>>>>>>>>>>> my point of view, pure streaming should be
>> the most resource
>> >>>>>>>>>>>>>>> demanding
>> >>>>>>>>>>>>>>> (if not, why bother with batch? why not run
>> everything in
>> >>>>>>>>>>>>>>> streaming
>> >>>>>>>>>>>>>>> only? what will there remain to "unify"?).
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Fortunately, for batch, only the state
>> for a single key
>> >>>>>>>>>>>>>>> needs to be
>> >>>>>>>>>>>>>>> preserved at a time, rather than the state
>> for all keys across
>> >>>>>>>>>>>>>>> the range
>> >>>>>>>>>>>>>>> of skew. Of course if you have few or hot
>> keys, one can still
>> >>>>>>>>>>>>>>> have
>> >>>>>>>>>>>>>>> issues (and this is not specific to
>> StatefulDoFns).
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Yes, but here is still the presumption that
>> my stateful DoFn can
>> >>>>>>>>>>>>>>> tolerate arbitrary shuffling of inputs. Let
>> me explain the use
>> >>>>>>>>>>>>>>> case in
>> >>>>>>>>>>>>>>> more detail.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Suppose you have input stream consisting of
>> 1s and 0s (and
>> >>>>>>>>>>>>>>> some key for
>> >>>>>>>>>>>>>>> each element, which is irrelevant for the
>> demonstration). Your
>> >>>>>>>>>>>>>>> task is
>> >>>>>>>>>>>>>>> to calculate in running global window the
>> actual number of
>> >>>>>>>>>>>>>>> changes
>> >>>>>>>>>>>>>>> between state 0 and state 1 and vice versa.
>> When the state
>> >>>>>>>>>>>>>>> doesn't
>> >>>>>>>>>>>>>>> change, you don't calculate anything. If
>> input (for given key)
>> >>>>>>>>>>>>>>> would be
>> >>>>>>>>>>>>>>> (tN denotes timestamp N):
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t1: 1
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t2: 0
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t3: 0
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t4: 1
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t5: 1
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t6: 0
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> then the output should yield (supposing that
>> default state is
>> >>>>>>>>>>>>>>> zero):
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t1: (one: 1, zero: 0)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t2: (one: 1, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t3: (one: 1, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t4: (one: 2, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t5: (one: 2, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t6: (one: 2, zero: 2)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> How would you implement this in current Beam
>> semantics?
>> >>>>>>>>>>>>>> I think your saying here that I know that my
>> input is ordered
>> >>>>>>>>>>>>>> in a specific way and since I assume the order
>> when writing my
>> >>>>>>>>>>>>>> pipeline I can perform this optimization. But
>> there is nothing
>> >>>>>>>>>>>>>> preventing a runner from noticing that your
>> processing in the
>> >>>>>>>>>>>>>> global window with a specific type of trigger
>> and re-ordering
>> >>>>>>>>>>>>>> your inputs/processing to get better
>> performance (since you
>> >>>>>>>>>>>>>> can't use an AfterWatermark trigger for your
>> pipeline in
>> >>>>>>>>>>>>>> streaming for the GlobalWindow).
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> Today, if you must have a strict order, you
>> must guarantee that
>> >>>>>>>>>>>>>> your StatefulParDo implements the necessary
>> "buffering &
>> >>>>>>>>>>>>>> sorting" into state. I can see why you would
>> want an annotation
>> >>>>>>>>>>>>>> that says I must have timestamp ordered
>> elements, since it
>> >>>>>>>>>>>>>> makes writing certain StatefulParDos much
>> easier. StatefulParDo
>> >>>>>>>>>>>>>> is a low-level function, it really is the
>> "here you go and do
>> >>>>>>>>>>>>>> whatever you need to but here be dragons"
>> function while
>> >>>>>>>>>>>>>> windowing and triggering is meant to keep many
>> people from
>> >>>>>>>>>>>>>> writing StatefulParDo in the first place.
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Pipelines that fail in the "worst case"
>> batch scenario
>> >>>>>>>>>>>>>>> are likely to
>> >>>>>>>>>>>>>>> degrade poorly (possibly catastrophically)
>> when the watermark
>> >>>>>>>>>>>>>>> falls
>> >>>>>>>>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> But the worst case is defined by input of
>> size (available
>> >>>>>>>>>>>>>>> resources +
>> >>>>>>>>>>>>>>> single byte) -> pipeline fail. Although it
>> could have
>> >>>>>>>>>>>>>>> finished, given
>> >>>>>>>>>>>>>>> the right conditions.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > This might be reasonable, implemented
>> by default by
>> >>>>>>>>>>>>>>> buffering
>> >>>>>>>>>>>>>>> everything and releasing elements as the
>> watermark (+lateness)
>> >>>>>>>>>>>>>>> advances,
>> >>>>>>>>>>>>>>> but would likely lead to inefficient (though
>> *maybe* easier to
>> >>>>>>>>>>>>>>> reason
>> >>>>>>>>>>>>>>> about) code.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Sure, the pipeline will be less efficient,
>> because it would
>> >>>>>>>>>>>>>>> have to
>> >>>>>>>>>>>>>>> buffer and sort the inputs. But at least it
>> will produce
>> >>>>>>>>>>>>>>> correct results
>> >>>>>>>>>>>>>>> in cases where updates to state are
>> order-sensitive.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Would it be roughly equivalent to GBK +
>> FlatMap(lambda
>> >>>>>>>>>>>>>>> (key, values):
>> >>>>>>>>>>>>>>> [(key, value) for value in values])?
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> I'd say roughly yes, but difference would be
>> in the trigger.
>> >>>>>>>>>>>>>>> The trigger
>> >>>>>>>>>>>>>>> should ideally fire as soon as watermark
>> (+lateness) crosses
>> >>>>>>>>>>>>>>> element
>> >>>>>>>>>>>>>>> with lowest timestamp in the buffer. Although
>> this could be
>> >>>>>>>>>>>>>>> somehow
>> >>>>>>>>>>>>>>> emulated by fixed trigger each X millis.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Or is the underlying desire just to be
>> able to hint to
>> >>>>>>>>>>>>>>> the runner
>> >>>>>>>>>>>>>>> that the code may perform better (e.g.
>> require less resources)
>> >>>>>>>>>>>>>>> as skew
>> >>>>>>>>>>>>>>> is reduced (and hence to order by timestamp
>> iff it's cheap)?
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> No, the sorting would have to be done in
>> streaming case as
>> >>>>>>>>>>>>>>> well. That is
>> >>>>>>>>>>>>>>> an imperative of the unified model. I think
>> it is possible to
>> >>>>>>>>>>>>>>> sort by
>> >>>>>>>>>>>>>>> timestamp only in batch case (and do it for
>> *all* batch
>> >>>>>>>>>>>>>>> stateful pardos
>> >>>>>>>>>>>>>>> without annotation), or introduce annotation,
>> but then make
>> >>>>>>>>>>>>>>> the same
>> >>>>>>>>>>>>>>> guarantees for streaming case as well.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Jan
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>> >>>>>>>>>>>>>>>> <je.ik@seznam.cz <ma...@seznam.cz>>
>> wrote:
>> >>>>>>>>>>>>>>>>> Hi Robert,
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> yes, I think you rephrased my point -
>> although no *explicit*
>> >>>>>>>>>>>>>>>>> guarantees
>> >>>>>>>>>>>>>>>>> of ordering are given in either mode, there
>> is *implicit*
>> >>>>>>>>>>>>>>>>> ordering in
>> >>>>>>>>>>>>>>>>> streaming case that is due to nature of the
>> processing - the
>> >>>>>>>>>>>>>>>>> difference
>> >>>>>>>>>>>>>>>>> between watermark and timestamp of elements
>> flowing through
>> >>>>>>>>>>>>>>>>> the pipeline
>> >>>>>>>>>>>>>>>>> are generally low (too high difference
>> leads to the
>> >>>>>>>>>>>>>>>>> overbuffering
>> >>>>>>>>>>>>>>>>> problem), but there is no such bound on batch.
>> >>>>>>>>>>>>>>>> Fortunately, for batch, only the state for a
>> single key needs
>> >>>>>>>>>>>>>>>> to be
>> >>>>>>>>>>>>>>>> preserved at a time, rather than the state
>> for all keys
>> >>>>>>>>>>>>>>>> across the
>> >>>>>>>>>>>>>>>> range of skew. Of course if you have few or
>> hot keys, one can
>> >>>>>>>>>>>>>>>> still
>> >>>>>>>>>>>>>>>> have issues (and this is not specific to
>> StatefulDoFns).
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> As a result, I see a few possible solutions:
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> - the best and most natural seems to be
>> extension of
>> >>>>>>>>>>>>>>>>> the model, so
>> >>>>>>>>>>>>>>>>> that it defines batch as not only
>> "streaming pipeline
>> >>>>>>>>>>>>>>>>> executed in batch
>> >>>>>>>>>>>>>>>>> fashion", but "pipeline with at least as
>> good runtime
>> >>>>>>>>>>>>>>>>> characteristics as
>> >>>>>>>>>>>>>>>>> in streaming case, executed in batch
>> fashion", I really
>> >>>>>>>>>>>>>>>>> don't think that
>> >>>>>>>>>>>>>>>>> there are any conflicts with the current
>> model, or that this
>> >>>>>>>>>>>>>>>>> could
>> >>>>>>>>>>>>>>>>> affect performance, because the required
>> sorting (as pointed by
>> >>>>>>>>>>>>>>>>> Aljoscha) is very probably already done
>> during translation
>> >>>>>>>>>>>>>>>>> of stateful
>> >>>>>>>>>>>>>>>>> pardos. Also note that this definition only
>> affects user
>> >>>>>>>>>>>>>>>>> defined
>> >>>>>>>>>>>>>>>>> stateful pardos
>> >>>>>>>>>>>>>>>> I don't see batch vs. streaming as part of
>> the model. One can
>> >>>>>>>>>>>>>>>> have
>> >>>>>>>>>>>>>>>> microbatch, or even a runner that alternates
>> between
>> >>>>>>>>>>>>>>>> different modes.
>> >>>>>>>>>>>>>>>> The model describes what the valid outputs
>> are given a
>> >>>>>>>>>>>>>>>> (sometimes
>> >>>>>>>>>>>>>>>> partial) set of inputs. It becomes really
>> hard to define
>> >>>>>>>>>>>>>>>> things like
>> >>>>>>>>>>>>>>>> "as good runtime characteristics." Once you
>> allow any
>> >>>>>>>>>>>>>>>> out-of-orderedness, it is not very feasible
>> to try and define
>> >>>>>>>>>>>>>>>> (and
>> >>>>>>>>>>>>>>>> more cheaply implement) a "upper bound" of
>> acceptable
>> >>>>>>>>>>>>>>>> out-of-orderedness.
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>> Pipelines that fail in the "worst case"
>> batch scenario are
>> >>>>>>>>>>>>>>>> likely to
>> >>>>>>>>>>>>>>>> degrade poorly (possibly catastrophically)
>> when the watermark
>> >>>>>>>>>>>>>>>> falls
>> >>>>>>>>>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> - another option would be to introduce
>> annotation for
>> >>>>>>>>>>>>>>>>> DoFns (e.g.
>> >>>>>>>>>>>>>>>>> @RequiresStableTimeCharacteristics), which
>> would result in
>> >>>>>>>>>>>>>>>>> the sorting
>> >>>>>>>>>>>>>>>>> in batch case - but - this extension would
>> have to ensure
>> >>>>>>>>>>>>>>>>> the sorting in
>> >>>>>>>>>>>>>>>>> streaming mode also - it would require
>> definition of allowed
>> >>>>>>>>>>>>>>>>> lateness,
>> >>>>>>>>>>>>>>>>> and triggger (essentially similar to window)
>> >>>>>>>>>>>>>>>> This might be reasonable, implemented by
>> default by buffering
>> >>>>>>>>>>>>>>>> everything and releasing elements as the
>> watermark (+lateness)
>> >>>>>>>>>>>>>>>> advances, but would likely lead to
>> inefficient (though
>> >>>>>>>>>>>>>>>> *maybe* easier
>> >>>>>>>>>>>>>>>> to reason about) code. Not sure about the
>> semantics of
>> >>>>>>>>>>>>>>>> triggering
>> >>>>>>>>>>>>>>>> here, especially data-driven triggers. Would
>> it be roughly
>> >>>>>>>>>>>>>>>> equivalent
>> >>>>>>>>>>>>>>>> to GBK + FlatMap(lambda (key, values):
>> [(key, value) for
>> >>>>>>>>>>>>>>>> value in
>> >>>>>>>>>>>>>>>> values])?
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>> Or is the underlying desire just to be able
>> to hint to the
>> >>>>>>>>>>>>>>>> runner that
>> >>>>>>>>>>>>>>>> the code may perform better (e.g. require
>> less resources) as
>> >>>>>>>>>>>>>>>> skew is
>> >>>>>>>>>>>>>>>> reduced (and hence to order by timestamp iff
>> it's cheap)?
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> - last option would be to introduce
>> these "higher order
>> >>>>>>>>>>>>>>>>> guarantees" in
>> >>>>>>>>>>>>>>>>> some extension DSL (e.g. Euphoria), but
>> that seems to be the
>> >>>>>>>>>>>>>>>>> worst
>> >>>>>>>>>>>>>>>>> option to me
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> I see the first two options quite equally
>> good, although the
>> >>>>>>>>>>>>>>>>> letter one
>> >>>>>>>>>>>>>>>>> is probably more time consuming to
>> implement. But it would
>> >>>>>>>>>>>>>>>>> bring
>> >>>>>>>>>>>>>>>>> additional feature to streaming case as well.
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> Thanks for any thoughts.
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> Jan
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>> >>>>>>>>>>>>>>>>>> <je.ik@seznam.cz <ma...@seznam.cz>>
>> wrote:
>> >>>>>>>>>>>>>>>>>>> Hi Reuven,
>> >>>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just
>> fine in batch
>> >>>>>>>>>>>>>>>>>>>> runners.
>> >>>>>>>>>>>>>>>>>>> Stateful ParDo works in batch as far, as
>> the logic inside
>> >>>>>>>>>>>>>>>>>>> the state works for absolutely unbounded
>> out-of-orderness
>> >>>>>>>>>>>>>>>>>>> of elements. That basically (practically)
>> can work only
>> >>>>>>>>>>>>>>>>>>> for cases, where the order of input
>> elements doesn't
>> >>>>>>>>>>>>>>>>>>> matter. But, "state" can refer to "state
>> machine", and any
>> >>>>>>>>>>>>>>>>>>> time you have a state machine involved,
>> then the ordering
>> >>>>>>>>>>>>>>>>>>> of elements would matter.
>> >>>>>>>>>>>>>>>>>> No guarantees on order are provided in
>> *either* streaming
>> >>>>>>>>>>>>>>>>>> or batch
>> >>>>>>>>>>>>>>>>>> mode by the model. However, it is the case
>> that in order to
>> >>>>>>>>>>>>>>>>>> make
>> >>>>>>>>>>>>>>>>>> forward progress most streaming runners
>> attempt to limit
>> >>>>>>>>>>>>>>>>>> the amount of
>> >>>>>>>>>>>>>>>>>> out-of-orderedness of elements (in terms
>> of event time vs.
>> >>>>>>>>>>>>>>>>>> processing
>> >>>>>>>>>>>>>>>>>> time) to make forward progress, which in
>> turn could help
>> >>>>>>>>>>>>>>>>>> cap the
>> >>>>>>>>>>>>>>>>>> amount of state that must be held
>> concurrently, whereas a
>> >>>>>>>>>>>>>>>>>> batch runner
>> >>>>>>>>>>>>>>>>>> may not allow any state to be safely
>> discarded until the whole
>> >>>>>>>>>>>>>>>>>> timeline from infinite past to infinite
>> future has been
>> >>>>>>>>>>>>>>>>>> observed.
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>> Also, as pointed out, state is not
>> preserved "batch to
>> >>>>>>>>>>>>>>>>>> batch" in batch mode.
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian
>> Michels
>> >>>>>>>>>>>>>>>>>> <mxm@apache.org <ma...@apache.org>>
>> wrote:
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>>>> batch semantics and streaming
>> semantics differs only
>> >>>>>>>>>>>>>>>>>>>> in that I can have GlobalWindow with
>> default trigger on
>> >>>>>>>>>>>>>>>>>>>> batch and cannot on stream
>> >>>>>>>>>>>>>>>>>>> You can have a GlobalWindow in streaming
>> with a default
>> >>>>>>>>>>>>>>>>>>> trigger. You
>> >>>>>>>>>>>>>>>>>>> could define additional triggers that do
>> early firings.
>> >>>>>>>>>>>>>>>>>>> And you could
>> >>>>>>>>>>>>>>>>>>> even trigger the global window by
>> advancing the watermark
>> >>>>>>>>>>>>>>>>>>> to +inf.
>> >>>>>>>>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited
>> global window with
>> >>>>>>>>>>>>>>>>>> default
>> >>>>>>>>>>>>>>>>>> trigger on unbounded PCollections in the
>> SDK because this
>> >>>>>>>>>>>>>>>>>> is more
>> >>>>>>>>>>>>>>>>>> likely to be user error than an actual
>> desire to have no
>> >>>>>>>>>>>>>>>>>> output until
>> >>>>>>>>>>>>>>>>>> drain. But it's semantically valid in the
>> model.
>>
Re: Definition of Unified model
Posted by Reuven Lax <re...@google.com>.
A slightly larger concern: it also will force users to create stateful
DoFns everywhere to generate these sequence numbers. If I have a ParDo that
is not a simple 1:1 transform (i.e. not MapElements), then the ParDo will
need to generate its own sequence numbers for ordering, and the only safe
way to do so is to use a stateful DoFn. This turns what used to be a simple
in-memory DoFn into one that has to access state. Also I believe many
runners will not fuse stateful DoFns. While none of this poses a problem
for the model, it could make ordering extremely expensive to achieve.
Reuven
On Tue, May 28, 2019 at 6:09 AM Jan Lukavský <je...@seznam.cz> wrote:
> Hi Reuven,
>
> > It also gets awkward with Flatten - the sequence number is no longer
> enough, you must also encode which side of the flatten each element came
> from.
>
> That is a generic need. Even if you read data from Kafka, the offsets are
> comparable only inside single partition. So, for Kafka to work as a FIFO
> for ordering, elements with same key have to be pushed to the same
> partition (otherwise Kafka cannot act as FIFO, because different partitions
> can be handled by different brokers, which means different observers and
> they therefore might not agree on the order of events). So if we want to
> emulate FIFO per key, then the sequence IDs have also be per key.
> On 5/28/19 2:33 PM, Reuven Lax wrote:
>
> Sequence metadata does have the disadvantage that users can no longer use
> the types coming from the source. You must create a new type that contains
> a sequence number (unless Beam provides this). It also gets awkward with
> Flatten - the sequence number is no longer enough, you must also encode
> which side of the flatten each element came from.
>
> On Tue, May 28, 2019 at 3:18 AM Jan Lukavský <je...@seznam.cz> wrote:
>
>> As I understood it, Kenn was supporting the idea that sequence metadata
>> is preferable over FIFO. I was trying to point out, that it even should
>> provide the same functionally as FIFO, plus one important more -
>> reproducibility and ability to being persisted and reused the same way
>> in batch and streaming.
>>
>> There is no doubt, that sequence metadata can be stored in every
>> storage. But, regarding some implicit ordering that sources might have -
>> yes, of course, data written into HDFS or Cloud Storage has ordering,
>> but only partial - inside some bulk (e.g. file) and the ordering is not
>> defined correctly on boundaries of these bulks (between files). That is
>> why I'd say, that ordering of sources is relevant only for
>> (partitioned!) streaming sources and generally always reduces to
>> sequence metadata (e.g. offsets).
>>
>> Jan
>>
>> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
>> > Huge +1 to all Kenn said.
>> >
>> > Jan, batch sources can have orderings too, just like Kafka. I think
>> > it's reasonable (for both batch and streaming) that if a source has an
>> > ordering that is an important part of the data, it should preserve
>> > this ordering into the data itself (e.g. as sequence numbers, offsets,
>> > etc.)
>> >
>> > On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org>
>> wrote:
>> >> I strongly prefer explicit sequence metadata over FIFO requirements,
>> because:
>> >>
>> >> - FIFO is complex to specify: for example Dataflow has "per stage
>> key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage"
>> is not a portable concept, nor even guaranteed to remain a Dataflow concept)
>> >> - complex specifications are by definition poor usability (if
>> necessary, then it is what it is)
>> >> - overly restricts the runner, reduces parallelism, for example any
>> non-stateful ParDo has per-element parallelism, not per "key"
>> >> - another perspective on that: FIFO makes everyone pay rather than
>> just the transform that requires exactly sequencing
>> >> - previous implementation details like reshuffles become part of the
>> model
>> >> - I'm not even convinced the use cases involved are addressed by
>> some careful FIFO restrictions; many sinks re-key and they would all have
>> to become aware of how keying of a sequence of "stages" affects the
>> end-to-end FIFO
>> >>
>> >> A noop becoming a non-noop is essentially the mathematical definition
>> of moving from higher-level to lower-level abstraction.
>> >>
>> >> So this strikes at the core question of what level of abstraction Beam
>> aims to represent. Lower-level means there are fewer possible
>> implementations and it is more tied to the underlying architecture, and
>> anything not near-exact match pays a huge penalty. Higher-level means there
>> are more implementations possible with different tradeoffs, though they may
>> all pay a minor penalty.
>> >>
>> >> I could be convinced to change my mind, but it needs some extensive
>> design, examples, etc. I think it is probably about the most consequential
>> design decision in the whole Beam model, around the same level as the
>> decision to use ParDo and GBK as the primitives IMO.
>> >>
>> >> Kenn
>> >>
>> >> On Thu, May 23, 2019 at 10:17 AM Reuven Lax <re...@google.com> wrote:
>> >>> Not really. I'm suggesting that some variant of FIFO ordering is
>> necessary, which requires either runners natively support FIFO ordering or
>> transforms adding some extra sequence number to each record to sort by.
>> >>>
>> >>> I still think your proposal is very useful by the way. I'm merely
>> pointing out that to solve the state-machine problem we probably need
>> something more.
>> >>>
>> >>> Reuven
>> >>>
>> >>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je...@seznam.cz> wrote:
>> >>>> Hi,
>> >>>> yes. It seems that ordering by user supplied UDF makes sense and I
>> will update the design proposal accordingly.
>> >>>> Would that solve the issues you mention?
>> >>>> Jan
>> >>>> ---------- Původní e-mail ----------
>> >>>> Od: Reuven Lax <re...@google.com>
>> >>>> Komu: dev <de...@beam.apache.org>
>> >>>> Datum: 23. 5. 2019 18:44:38
>> >>>> Předmět: Re: Definition of Unified model
>> >>>>
>> >>>> I'm simply saying that timestamp ordering is insufficient for state
>> machines. I wasn't proposing Kafka as a solution - that was simply an
>> example of how people solve this problem in other scenarios.
>> >>>>
>> >>>> BTW another example of ordering: Imagine today that you have a
>> triggered Sum aggregation writing out to a key-value sink. In theory we
>> provide no ordering, so the sink might write the triggered sums in the
>> wrong order, ending up with an incorrect value in the sink. In this case
>> you probably want values ordered by trigger pane index.
>> >>>>
>> >>>> Reuven
>> >>>>
>> >>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz>
>> wrote:
>> >>>>
>> >>>> Hi Reuven,
>> >>>> I share the view point of Robert. I think the isuue you refer to is
>> not in reality related to timestamps, but to the fact, that ordering of
>> events in time is observer dependent (either caused by relativity, or time
>> skew, essentially this has the same consequences). And the resolution in
>> fact isn't Kafka, but generally an authoritative observer, that tells you
>> "I saw the events in this order". And you either have one (and have the
>> outcome of his observation persisted in the data - e.g. as offset in Kafka
>> partition), then you should be able to use it (maybe that suggests afterall
>> that sorting by some user supplied UDF might make sense), or do not have
>> it, and then any interpretation of the data seems to be equally valid.
>> Although determinism is fine, of course.
>> >>>> Jan
>> >>>> ---------- Původní e-mail ----------
>> >>>> Od: Reuven Lax <re...@google.com>
>> >>>> Komu: dev <de...@beam.apache.org>
>> >>>> Datum: 23. 5. 2019 17:39:12
>> >>>> Předmět: Re: Definition of Unified model
>> >>>>
>> >>>> So an example would be elements of type "startUserSession" and
>> "endUserSession" (website sessions, not Beam sessions). Logically you may
>> need to process them in the correct order if you have any sort of
>> state-machine logic. However timestamp ordering is never guaranteed to
>> match the logical ordering. Not only might you have several elements with
>> the same timestamp, but in reality time skew across backend servers can
>> cause the events to have timestamps in reverse order of the actual
>> causality order.
>> >>>>
>> >>>> People do solve this problem today though. Publish the events to
>> Kafka, making sure that events for the same user end up in the same Kafka
>> partition. This ensures that the events appear in the Kafka partitions in
>> causality order, even if the timestamp order doesn't match. The your Kafka
>> subscriber simply process the elements in each partition in order.
>> >>>>
>> >>>> I think the ability to impose FIFO causality ordering is what's
>> needed for any state-machine work. Timestamp ordering has advantages
>> (though often I think the advantage is in state), but does not solve this
>> problem.
>> >>>>
>> >>>> Reuven
>> >>>>
>> >>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <ro...@google.com>
>> wrote:
>> >>>>
>> >>>> Good point.
>> >>>>
>> >>>> The "implementation-specific" way I would do this is
>> >>>> window-by-instant, followed by a DoFn that gets all the elements with
>> >>>> the same timestamp and sorts/acts accordingly, but this counts on the
>> >>>> runner producing windows in timestamp order (likely?) and also the
>> >>>> subsequent DoFn getting them in this order (also likely, due to
>> >>>> fusion).
>> >>>>
>> >>>> One could make the argument that, though it does not provide
>> >>>> deterministic behavior, getting elements of the same timestamp in
>> >>>> different orders should produce equally valid interpretations of the
>> >>>> data. (After all, due to relatively, timestamps are not technically
>> >>>> well ordered across space.) I can see how data-dependent tiebreakers
>> >>>> could be useful, or promises of preservation of order between
>> >>>> operations.
>> >>>>
>> >>>> - Robert
>> >>>>
>> >>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com> wrote:
>> >>>>> So Jan's example of state machines is quite a valid use case for
>> ordering. However in my experience, timestamp ordering is insufficient for
>> state machines. Elements that cause state transitions might come in with
>> the exact same timestamp, yet still have a necessary ordering. Especially
>> given Beam's decision to have milliseconds timestamps this is possible, but
>> even at microsecond or nanosecond precision this can happen at scale. To
>> handle state machines you usually need some sort of FIFO ordering along
>> with an ordered sources, such as Kafka, not timestamp ordering.
>> >>>>>
>> >>>>> Reuven
>> >>>>>
>> >>>>> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz>
>> wrote:
>> >>>>>> Hi all,
>> >>>>>>
>> >>>>>> thanks everyone for this discussion. I think I have gathered enough
>> >>>>>> feedback to be able to put down a proposition for changes, which I
>> will
>> >>>>>> do and send to this list for further discussion. There are still
>> doubts
>> >>>>>> remaining the non-determinism and it's relation to outputs
>> stability vs.
>> >>>>>> latency. But I will try to clarify all this in the design document.
>> >>>>>>
>> >>>>>> Thanks,
>> >>>>>>
>> >>>>>> Jan
>> >>>>>>
>> >>>>>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
>> >>>>>>>> current understanding.
>> >>>>>>> In essence your description of how exactly-once works in Flink is
>> >>>>>>> correct. The general assumption in Flink is that pipelines must be
>> >>>>>>> deterministic and thus produce idempotent writes in the case of
>> >>>>>>> failures. However, that doesn't mean Beam sinks can't guarantee a
>> bit
>> >>>>>>> more with what Flink has to offer.
>> >>>>>>>
>> >>>>>>> Luke already mentioned the design discussions for
>> @RequiresStableInput
>> >>>>>>> which ensures idempotent writes for non-deterministic pipelines.
>> This
>> >>>>>>> is not part of the model but an optional Beam feature.
>> >>>>>>>
>> >>>>>>> We recently implemented support for @RequiresStableInput in the
>> Flink
>> >>>>>>> Runner. Reuven mentioned the Flink checkpoint confirmation, which
>> >>>>>>> allows us to buffer (and checkpoint) processed data and only emit
>> it
>> >>>>>>> once a Flink checkpoint has completed.
>> >>>>>>>
>> >>>>>>> Cheers,
>> >>>>>>> Max
>> >>>>>>>
>> >>>>>>> On 21.05.19 16:49, Jan Lukavský wrote:
>> >>>>>>>> Hi,
>> >>>>>>>>
>> >>>>>>>> > Actually, I think it is a larger (open) question whether
>> exactly
>> >>>>>>>> once is guaranteed by the model or whether runners are allowed to
>> >>>>>>>> relax that. I would think, however, that sources correctly
>> >>>>>>>> implemented should be idempotent when run atop an exactly once
>> >>>>>>>> infrastructure such as Flink of Dataflow.
>> >>>>>>>>
>> >>>>>>>> I would assume, that the model basically inherits guarantees of
>> >>>>>>>> underlying infrastructure. Because Flink does not work as you
>> >>>>>>>> described (atomic commit of inputs, state and outputs), but
>> rather a
>> >>>>>>>> checkpoint mark is flowing through the DAG much like watermark
>> and on
>> >>>>>>>> failures operators are restored and data reprocessed, it (IMHO)
>> >>>>>>>> implies, that you have exactly once everywhere in the DAG *but*
>> >>>>>>>> sinks. That is because sinks cannot be restored to previous
>> state,
>> >>>>>>>> instead sinks are supposed to be idempotent in order for the
>> exactly
>> >>>>>>>> once to really work (or at least be able to commit outputs on
>> >>>>>>>> checkpoint in sink). That implies that if you don't have sink
>> that is
>> >>>>>>>> able to commit outputs atomically on checkpoint, the pipeline
>> >>>>>>>> execution should be deterministic upon retries, otherwise shadow
>> >>>>>>>> writes from failed paths of the pipeline might appear.
>> >>>>>>>>
>> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
>> >>>>>>>> current understanding.
>> >>>>>>>>
>> >>>>>>>> > Sounds like we should make this clearer.
>> >>>>>>>>
>> >>>>>>>> I meant that you are right that we must not in any thoughts we
>> are
>> >>>>>>>> having forget that streams are by definition out-of-order. That
>> is
>> >>>>>>>> property that we cannot change. But - that doesn't limit us from
>> >>>>>>>> creating operator that presents the data to UDF as if the stream
>> was
>> >>>>>>>> ideally sorted. It can do that by introducing latency, of course.
>> >>>>>>>>
>> >>>>>>>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>> >>>>>>>>> Reza: One could provide something like this as a utility class,
>> but
>> >>>>>>>>> one downside is that it is not scale invariant. It requires a
>> tuning
>> >>>>>>>>> parameter that, if to small, won't mitigate the problem, but if
>> to
>> >>>>>>>>> big, greatly increases latency. (Possibly one could define a
>> dynamic
>> >>>>>>>>> session-like window to solve this though...) It also might be
>> harder
>> >>>>>>>>> for runners that *can* cheaply present stuff in timestamp order
>> to
>> >>>>>>>>> optimize. (That and, in practice, our annotation-style process
>> methods
>> >>>>>>>>> don't lend themselves to easy composition.) I think it could
>> work in
>> >>>>>>>>> specific cases though.
>> >>>>>>>>>
>> >>>>>>>>> More inline below.
>> >>>>>>>>>
>> >>>>>>>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz>
>> wrote:
>> >>>>>>>>>> Hi Robert,
>> >>>>>>>>>>
>> >>>>>>>>>> > Beam has an exactly-once model. If the data was consumed,
>> state
>> >>>>>>>>>> mutated, and outputs written downstream (these three are
>> committed
>> >>>>>>>>>> together atomically) it will not be replayed. That does not, of
>> >>>>>>>>>> course,
>> >>>>>>>>>> solve the non-determanism due to ordering (including the fact
>> that two
>> >>>>>>>>>> operations reading the same PCollection may view different
>> ordering).
>> >>>>>>>>>>
>> >>>>>>>>>> I think what you describe is a property of a runner, not of
>> the model,
>> >>>>>>>>>> right? I think if I run my pipeline on Flink I will not get
>> this
>> >>>>>>>>>> atomicity, because although Flink uses also exactly-once model
>> if
>> >>>>>>>>>> might
>> >>>>>>>>>> write outputs multiple times.
>> >>>>>>>>> Actually, I think it is a larger (open) question whether
>> exactly once
>> >>>>>>>>> is guaranteed by the model or whether runners are allowed to
>> relax
>> >>>>>>>>> that. I would think, however, that sources correctly implemented
>> >>>>>>>>> should be idempotent when run atop an exactly once
>> infrastructure such
>> >>>>>>>>> as Flink of Dataflow.
>> >>>>>>>>>
>> >>>>>>>>>> > 1) Is it correct for a (Stateful)DoFn to assume elements
>> are
>> >>>>>>>>>> received
>> >>>>>>>>>> in a specific order? In the current model, it is not. Being
>> able to
>> >>>>>>>>>> read, handle, and produced out-of-order data, including late
>> data,
>> >>>>>>>>>> is a
>> >>>>>>>>>> pretty fundamental property of distributed systems.
>> >>>>>>>>>>
>> >>>>>>>>>> Yes, absolutely. The argument here is not that Stateful ParDo
>> should
>> >>>>>>>>>> presume to receive elements in any order, but to _present_ it
>> as
>> >>>>>>>>>> such to
>> >>>>>>>>>> the user @ProcessElement function.
>> >>>>>>>>> Sounds like we should make this clearer.
>> >>>>>>>>>
>> >>>>>>>>>> > 2) Given that some operations are easier (or possibly only
>> >>>>>>>>>> possible)
>> >>>>>>>>>> to write when operating on ordered data, and that different
>> runners
>> >>>>>>>>>> may
>> >>>>>>>>>> have (significantly) cheaper ways to provide this ordering
>> than can be
>> >>>>>>>>>> done by the user themselves, should we elevate this to a
>> property of
>> >>>>>>>>>> (Stateful?)DoFns that the runner can provide? I think a
>> compelling
>> >>>>>>>>>> argument can be made here that we should.
>> >>>>>>>>>>
>> >>>>>>>>>> +1
>> >>>>>>>>>>
>> >>>>>>>>>> Jan
>> >>>>>>>>>>
>> >>>>>>>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>> >>>>>>>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz>
>> wrote:
>> >>>>>>>>>>>> > I don't see batch vs. streaming as part of the model.
>> One
>> >>>>>>>>>>>> can have
>> >>>>>>>>>>>> microbatch, or even a runner that alternates between
>> different
>> >>>>>>>>>>>> modes.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Although I understand motivation of this statement, this
>> project
>> >>>>>>>>>>>> name is
>> >>>>>>>>>>>> "Apache Beam: An advanced unified programming model". What
>> does the
>> >>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the
>> model?
>> >>>>>>>>>>> What I mean is that streaming vs. batch is no longer part of
>> the
>> >>>>>>>>>>> model
>> >>>>>>>>>>> (or ideally API), but pushed down to be a concern of the
>> runner
>> >>>>>>>>>>> (executor) of the pipeline.
>> >>>>>>>>>>>
>> >>>>>>>>>>>
>> >>>>>>>>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <
>> je.ik@seznam.cz>
>> >>>>>>>>>>> wrote:
>> >>>>>>>>>>>> Hi Kenn,
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> OK, so if we introduce annotation, we can have stateful ParDo
>> >>>>>>>>>>>> with sorting, that would perfectly resolve my issues. I still
>> >>>>>>>>>>>> have some doubts, though. Let me explain. The current
>> behavior of
>> >>>>>>>>>>>> stateful ParDo has the following properties:
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> a) might fail in batch, although runs fine in streaming
>> (that
>> >>>>>>>>>>>> is due to the buffering, and unbounded lateness in batch,
>> which
>> >>>>>>>>>>>> was discussed back and forth in this thread)
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> b) might be non deterministic (this is because the
>> elements
>> >>>>>>>>>>>> arrive at somewhat random order, and even if you do the
>> operation
>> >>>>>>>>>>>> "assign unique ID to elements" this might produce different
>> >>>>>>>>>>>> results when run multiple times)
>> >>>>>>>>>>> PCollections are *explicitly* unordered. Any operations that
>> >>>>>>>>>>> assume or
>> >>>>>>>>>>> depend on a specific ordering for correctness (or
>> determinism) must
>> >>>>>>>>>>> provide that ordering themselves (i.e. tolerate "arbitrary
>> shuffling
>> >>>>>>>>>>> of inputs"). As you point out, that may be very expensive if
>> you have
>> >>>>>>>>>>> very hot keys with very large (unbounded) timestamp skew.
>> >>>>>>>>>>>
>> >>>>>>>>>>> StatefulDoFns are low-level operations that should be used
>> with care;
>> >>>>>>>>>>> the simpler windowing model gives determinism in the face of
>> >>>>>>>>>>> unordered
>> >>>>>>>>>>> data (though late data and non-end-of-window triggering
>> introduces
>> >>>>>>>>>>> some of the non-determanism back in).
>> >>>>>>>>>>>
>> >>>>>>>>>>>> What worries me most is the property b), because it seems to
>> me
>> >>>>>>>>>>>> to have serious consequences - not only that if you run twice
>> >>>>>>>>>>>> batch pipeline you would get different results, but even on
>> >>>>>>>>>>>> streaming, when pipeline fails and gets restarted from
>> >>>>>>>>>>>> checkpoint, produced output might differ from the previous
>> run
>> >>>>>>>>>>>> and data from the first run might have already been persisted
>> >>>>>>>>>>>> into sink. That would create somewhat messy outputs.
>> >>>>>>>>>>> Beam has an exactly-once model. If the data was consumed,
>> state
>> >>>>>>>>>>> mutated, and outputs written downstream (these three are
>> committed
>> >>>>>>>>>>> together atomically) it will not be replayed. That does not,
>> of
>> >>>>>>>>>>> course, solve the non-determanism due to ordering (including
>> the fact
>> >>>>>>>>>>> that two operations reading the same PCollection may view
>> different
>> >>>>>>>>>>> ordering).
>> >>>>>>>>>>>
>> >>>>>>>>>>>> These two properties makes me think that the current
>> >>>>>>>>>>>> implementation is more of a _special case_ than the general
>> one.
>> >>>>>>>>>>>> The general one would be that your state doesn't have the
>> >>>>>>>>>>>> properties to be able to tolerate buffering problems and/or
>> >>>>>>>>>>>> non-determinism. Which is the case where you need sorting in
>> both
>> >>>>>>>>>>>> streaming and batch to be part of the model.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Let me point out one more analogy - that is merging vs.
>> >>>>>>>>>>>> non-merging windows. The general case (merging windows)
>> implies
>> >>>>>>>>>>>> sorting by timestamp in both batch case (explicit) and
>> streaming
>> >>>>>>>>>>>> (buffering). The special case (non-merging windows) doesn't
>> rely
>> >>>>>>>>>>>> on any timestamp ordering, so the sorting and buffering can
>> be
>> >>>>>>>>>>>> dropped. The underlying root cause of this is the same for
>> both
>> >>>>>>>>>>>> stateful ParDo and windowing (essentially, assigning window
>> >>>>>>>>>>>> labels is a stateful operation when windowing function is
>> merging).
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> The reason for the current behavior of stateful ParDo seems
>> to be
>> >>>>>>>>>>>> performance, but is it right to abandon correctness in favor
>> of
>> >>>>>>>>>>>> performance? Wouldn't it be more consistent to have the
>> default
>> >>>>>>>>>>>> behavior prefer correctness and when you have the specific
>> >>>>>>>>>>>> conditions of state function having special properties, then
>> you
>> >>>>>>>>>>>> can annotate your DoFn (with something like
>> >>>>>>>>>>>> @TimeOrderingAgnostic), which would yield a better
>> performance in
>> >>>>>>>>>>>> that case?
>> >>>>>>>>>>> There are two separable questions here.
>> >>>>>>>>>>>
>> >>>>>>>>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are
>> received
>> >>>>>>>>>>> in a specific order? In the current model, it is not. Being
>> able to
>> >>>>>>>>>>> read, handle, and produced out-of-order data, including late
>> data, is
>> >>>>>>>>>>> a pretty fundamental property of distributed systems.
>> >>>>>>>>>>>
>> >>>>>>>>>>> 2) Given that some operations are easier (or possibly only
>> possible)
>> >>>>>>>>>>> to write when operating on ordered data, and that different
>> runners
>> >>>>>>>>>>> may have (significantly) cheaper ways to provide this
>> ordering than
>> >>>>>>>>>>> can be done by the user themselves, should we elevate this to
>> a
>> >>>>>>>>>>> property of (Stateful?)DoFns that the runner can provide? I
>> think a
>> >>>>>>>>>>> compelling argument can be made here that we should.
>> >>>>>>>>>>>
>> >>>>>>>>>>> - Robert
>> >>>>>>>>>>>
>> >>>>>>>>>>>
>> >>>>>>>>>>>
>> >>>>>>>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Thanks for the nice small example of a calculation that
>> depends
>> >>>>>>>>>>>> on order. You are right that many state machines have this
>> >>>>>>>>>>>> property. I agree w/ you and Luke that it is convenient for
>> batch
>> >>>>>>>>>>>> processing to sort by event timestamp before running a
>> stateful
>> >>>>>>>>>>>> ParDo. In streaming you could also implement "sort by event
>> >>>>>>>>>>>> timestamp" by buffering until you know all earlier data will
>> be
>> >>>>>>>>>>>> dropped - a slack buffer up to allowed lateness.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> I do not think that it is OK to sort in batch and not in
>> >>>>>>>>>>>> streaming. Many state machines diverge very rapidly when
>> things
>> >>>>>>>>>>>> are out of order. So each runner if they see the
>> >>>>>>>>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
>> >>>>>>>>>>>> sorted data (by some mix of buffering and dropping), or to
>> reject
>> >>>>>>>>>>>> the pipeline as unsupported.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> And also want to say that this is not the default case - many
>> >>>>>>>>>>>> uses of state & timers in ParDo yield different results at
>> the
>> >>>>>>>>>>>> element level, but the results are equivalent at in the big
>> >>>>>>>>>>>> picture. Such as the example of "assign a unique sequence
>> number
>> >>>>>>>>>>>> to each element" or "group into batches" it doesn't matter
>> >>>>>>>>>>>> exactly what the result is, only that it meets the spec. And
>> >>>>>>>>>>>> other cases like user funnels are monotonic enough that you
>> also
>> >>>>>>>>>>>> don't actually need sorting.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Kenn
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <
>> je.ik@seznam.cz>
>> >>>>>>>>>>>> wrote:
>> >>>>>>>>>>>>> Yes, the problem will arise probably mostly when you have
>> not
>> >>>>>>>>>>>>> well distributed keys (or too few keys). I'm really not
>> sure if
>> >>>>>>>>>>>>> a pure GBK with a trigger can solve this - it might help to
>> have
>> >>>>>>>>>>>>> data driven trigger. There would still be some doubts,
>> though.
>> >>>>>>>>>>>>> The main question is still here - people say, that sorting
>> by
>> >>>>>>>>>>>>> timestamp before stateful ParDo would be prohibitively
>> slow, but
>> >>>>>>>>>>>>> I don't really see why - the sorting is very probably
>> already
>> >>>>>>>>>>>>> there. And if not (hash grouping instead of sorted
>> grouping),
>> >>>>>>>>>>>>> then the sorting would affect only user defined
>> StatefulParDos.
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> This would suggest that the best way out of this would be
>> really
>> >>>>>>>>>>>>> to add annotation, so that the author of the pipeline can
>> decide.
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> If that would be acceptable I think I can try to prepare
>> some
>> >>>>>>>>>>>>> basic functionality, but I'm not sure, if I would be able to
>> >>>>>>>>>>>>> cover all runners / sdks.
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> It is read all per key and window and not just read all
>> (this
>> >>>>>>>>>>>>> still won't scale with hot keys in the global window). The
>> GBK
>> >>>>>>>>>>>>> preceding the StatefulParDo will guarantee that you are
>> >>>>>>>>>>>>> processing all the values for a specific key and window at
>> any
>> >>>>>>>>>>>>> given time. Is there a specific window/trigger that is
>> missing
>> >>>>>>>>>>>>> that you feel would remove the need for you to use
>> StatefulParDo?
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <
>> je.ik@seznam.cz>
>> >>>>>>>>>>>>> wrote:
>> >>>>>>>>>>>>>> Hi Lukasz,
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Today, if you must have a strict order, you must guarantee
>> >>>>>>>>>>>>>>> that your StatefulParDo implements the necessary
>> "buffering &
>> >>>>>>>>>>>>>>> sorting" into state.
>> >>>>>>>>>>>>>> Yes, no problem with that. But this whole discussion
>> started,
>> >>>>>>>>>>>>>> because *this doesn't work on batch*. You simply cannot
>> first
>> >>>>>>>>>>>>>> read everything from distributed storage and then buffer
>> it all
>> >>>>>>>>>>>>>> into memory, just to read it again, but sorted. That will
>> not
>> >>>>>>>>>>>>>> work. And even if it would, it would be a terrible waste of
>> >>>>>>>>>>>>>> resources.
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> Jan
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <
>> je.ik@seznam.cz>
>> >>>>>>>>>>>>>> wrote:
>> >>>>>>>>>>>>>>> This discussion brings many really interesting questions
>> for
>> >>>>>>>>>>>>>>> me. :-)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > I don't see batch vs. streaming as part of the
>> model. One
>> >>>>>>>>>>>>>>> can have
>> >>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
>> different
>> >>>>>>>>>>>>>>> modes.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Although I understand motivation of this statement, this
>> >>>>>>>>>>>>>>> project name is
>> >>>>>>>>>>>>>>> "Apache Beam: An advanced unified programming model". What
>> >>>>>>>>>>>>>>> does the
>> >>>>>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the
>> model?
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Using microbatching, chaining of batch jobs, or pure
>> streaming
>> >>>>>>>>>>>>>>> are
>> >>>>>>>>>>>>>>> exactly the "runtime conditions/characteristics" I refer
>> to.
>> >>>>>>>>>>>>>>> All these
>> >>>>>>>>>>>>>>> define several runtime parameters, which in turn define
>> how
>> >>>>>>>>>>>>>>> well/badly
>> >>>>>>>>>>>>>>> will the pipeline perform and how many resources might be
>> >>>>>>>>>>>>>>> needed. From
>> >>>>>>>>>>>>>>> my point of view, pure streaming should be the most
>> resource
>> >>>>>>>>>>>>>>> demanding
>> >>>>>>>>>>>>>>> (if not, why bother with batch? why not run everything in
>> >>>>>>>>>>>>>>> streaming
>> >>>>>>>>>>>>>>> only? what will there remain to "unify"?).
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Fortunately, for batch, only the state for a single
>> key
>> >>>>>>>>>>>>>>> needs to be
>> >>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys
>> across
>> >>>>>>>>>>>>>>> the range
>> >>>>>>>>>>>>>>> of skew. Of course if you have few or hot keys, one can
>> still
>> >>>>>>>>>>>>>>> have
>> >>>>>>>>>>>>>>> issues (and this is not specific to StatefulDoFns).
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Yes, but here is still the presumption that my stateful
>> DoFn can
>> >>>>>>>>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain
>> the use
>> >>>>>>>>>>>>>>> case in
>> >>>>>>>>>>>>>>> more detail.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
>> >>>>>>>>>>>>>>> some key for
>> >>>>>>>>>>>>>>> each element, which is irrelevant for the demonstration).
>> Your
>> >>>>>>>>>>>>>>> task is
>> >>>>>>>>>>>>>>> to calculate in running global window the actual number of
>> >>>>>>>>>>>>>>> changes
>> >>>>>>>>>>>>>>> between state 0 and state 1 and vice versa. When the state
>> >>>>>>>>>>>>>>> doesn't
>> >>>>>>>>>>>>>>> change, you don't calculate anything. If input (for given
>> key)
>> >>>>>>>>>>>>>>> would be
>> >>>>>>>>>>>>>>> (tN denotes timestamp N):
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t1: 1
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t2: 0
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t3: 0
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t4: 1
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t5: 1
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t6: 0
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> then the output should yield (supposing that default
>> state is
>> >>>>>>>>>>>>>>> zero):
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t1: (one: 1, zero: 0)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t2: (one: 1, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t3: (one: 1, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t4: (one: 2, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t5: (one: 2, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t6: (one: 2, zero: 2)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> How would you implement this in current Beam semantics?
>> >>>>>>>>>>>>>> I think your saying here that I know that my input is
>> ordered
>> >>>>>>>>>>>>>> in a specific way and since I assume the order when
>> writing my
>> >>>>>>>>>>>>>> pipeline I can perform this optimization. But there is
>> nothing
>> >>>>>>>>>>>>>> preventing a runner from noticing that your processing in
>> the
>> >>>>>>>>>>>>>> global window with a specific type of trigger and
>> re-ordering
>> >>>>>>>>>>>>>> your inputs/processing to get better performance (since you
>> >>>>>>>>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>> >>>>>>>>>>>>>> streaming for the GlobalWindow).
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> Today, if you must have a strict order, you must guarantee
>> that
>> >>>>>>>>>>>>>> your StatefulParDo implements the necessary "buffering &
>> >>>>>>>>>>>>>> sorting" into state. I can see why you would want an
>> annotation
>> >>>>>>>>>>>>>> that says I must have timestamp ordered elements, since it
>> >>>>>>>>>>>>>> makes writing certain StatefulParDos much easier.
>> StatefulParDo
>> >>>>>>>>>>>>>> is a low-level function, it really is the "here you go and
>> do
>> >>>>>>>>>>>>>> whatever you need to but here be dragons" function while
>> >>>>>>>>>>>>>> windowing and triggering is meant to keep many people from
>> >>>>>>>>>>>>>> writing StatefulParDo in the first place.
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Pipelines that fail in the "worst case" batch
>> scenario
>> >>>>>>>>>>>>>>> are likely to
>> >>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the
>> watermark
>> >>>>>>>>>>>>>>> falls
>> >>>>>>>>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> But the worst case is defined by input of size (available
>> >>>>>>>>>>>>>>> resources +
>> >>>>>>>>>>>>>>> single byte) -> pipeline fail. Although it could have
>> >>>>>>>>>>>>>>> finished, given
>> >>>>>>>>>>>>>>> the right conditions.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > This might be reasonable, implemented by default by
>> >>>>>>>>>>>>>>> buffering
>> >>>>>>>>>>>>>>> everything and releasing elements as the watermark
>> (+lateness)
>> >>>>>>>>>>>>>>> advances,
>> >>>>>>>>>>>>>>> but would likely lead to inefficient (though *maybe*
>> easier to
>> >>>>>>>>>>>>>>> reason
>> >>>>>>>>>>>>>>> about) code.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Sure, the pipeline will be less efficient, because it
>> would
>> >>>>>>>>>>>>>>> have to
>> >>>>>>>>>>>>>>> buffer and sort the inputs. But at least it will produce
>> >>>>>>>>>>>>>>> correct results
>> >>>>>>>>>>>>>>> in cases where updates to state are order-sensitive.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Would it be roughly equivalent to GBK +
>> FlatMap(lambda
>> >>>>>>>>>>>>>>> (key, values):
>> >>>>>>>>>>>>>>> [(key, value) for value in values])?
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> I'd say roughly yes, but difference would be in the
>> trigger.
>> >>>>>>>>>>>>>>> The trigger
>> >>>>>>>>>>>>>>> should ideally fire as soon as watermark (+lateness)
>> crosses
>> >>>>>>>>>>>>>>> element
>> >>>>>>>>>>>>>>> with lowest timestamp in the buffer. Although this could
>> be
>> >>>>>>>>>>>>>>> somehow
>> >>>>>>>>>>>>>>> emulated by fixed trigger each X millis.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Or is the underlying desire just to be able to hint
>> to
>> >>>>>>>>>>>>>>> the runner
>> >>>>>>>>>>>>>>> that the code may perform better (e.g. require less
>> resources)
>> >>>>>>>>>>>>>>> as skew
>> >>>>>>>>>>>>>>> is reduced (and hence to order by timestamp iff it's
>> cheap)?
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> No, the sorting would have to be done in streaming case as
>> >>>>>>>>>>>>>>> well. That is
>> >>>>>>>>>>>>>>> an imperative of the unified model. I think it is
>> possible to
>> >>>>>>>>>>>>>>> sort by
>> >>>>>>>>>>>>>>> timestamp only in batch case (and do it for *all* batch
>> >>>>>>>>>>>>>>> stateful pardos
>> >>>>>>>>>>>>>>> without annotation), or introduce annotation, but then
>> make
>> >>>>>>>>>>>>>>> the same
>> >>>>>>>>>>>>>>> guarantees for streaming case as well.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Jan
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>> >>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>> >>>>>>>>>>>>>>>>> Hi Robert,
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> yes, I think you rephrased my point - although no
>> *explicit*
>> >>>>>>>>>>>>>>>>> guarantees
>> >>>>>>>>>>>>>>>>> of ordering are given in either mode, there is
>> *implicit*
>> >>>>>>>>>>>>>>>>> ordering in
>> >>>>>>>>>>>>>>>>> streaming case that is due to nature of the processing
>> - the
>> >>>>>>>>>>>>>>>>> difference
>> >>>>>>>>>>>>>>>>> between watermark and timestamp of elements flowing
>> through
>> >>>>>>>>>>>>>>>>> the pipeline
>> >>>>>>>>>>>>>>>>> are generally low (too high difference leads to the
>> >>>>>>>>>>>>>>>>> overbuffering
>> >>>>>>>>>>>>>>>>> problem), but there is no such bound on batch.
>> >>>>>>>>>>>>>>>> Fortunately, for batch, only the state for a single key
>> needs
>> >>>>>>>>>>>>>>>> to be
>> >>>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys
>> >>>>>>>>>>>>>>>> across the
>> >>>>>>>>>>>>>>>> range of skew. Of course if you have few or hot keys,
>> one can
>> >>>>>>>>>>>>>>>> still
>> >>>>>>>>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> As a result, I see a few possible solutions:
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> - the best and most natural seems to be extension
>> of
>> >>>>>>>>>>>>>>>>> the model, so
>> >>>>>>>>>>>>>>>>> that it defines batch as not only "streaming pipeline
>> >>>>>>>>>>>>>>>>> executed in batch
>> >>>>>>>>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>> >>>>>>>>>>>>>>>>> characteristics as
>> >>>>>>>>>>>>>>>>> in streaming case, executed in batch fashion", I really
>> >>>>>>>>>>>>>>>>> don't think that
>> >>>>>>>>>>>>>>>>> there are any conflicts with the current model, or that
>> this
>> >>>>>>>>>>>>>>>>> could
>> >>>>>>>>>>>>>>>>> affect performance, because the required sorting (as
>> pointed by
>> >>>>>>>>>>>>>>>>> Aljoscha) is very probably already done during
>> translation
>> >>>>>>>>>>>>>>>>> of stateful
>> >>>>>>>>>>>>>>>>> pardos. Also note that this definition only affects user
>> >>>>>>>>>>>>>>>>> defined
>> >>>>>>>>>>>>>>>>> stateful pardos
>> >>>>>>>>>>>>>>>> I don't see batch vs. streaming as part of the model.
>> One can
>> >>>>>>>>>>>>>>>> have
>> >>>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
>> >>>>>>>>>>>>>>>> different modes.
>> >>>>>>>>>>>>>>>> The model describes what the valid outputs are given a
>> >>>>>>>>>>>>>>>> (sometimes
>> >>>>>>>>>>>>>>>> partial) set of inputs. It becomes really hard to define
>> >>>>>>>>>>>>>>>> things like
>> >>>>>>>>>>>>>>>> "as good runtime characteristics." Once you allow any
>> >>>>>>>>>>>>>>>> out-of-orderedness, it is not very feasible to try and
>> define
>> >>>>>>>>>>>>>>>> (and
>> >>>>>>>>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>> >>>>>>>>>>>>>>>> out-of-orderedness.
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>> Pipelines that fail in the "worst case" batch scenario
>> are
>> >>>>>>>>>>>>>>>> likely to
>> >>>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the
>> watermark
>> >>>>>>>>>>>>>>>> falls
>> >>>>>>>>>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> - another option would be to introduce annotation
>> for
>> >>>>>>>>>>>>>>>>> DoFns (e.g.
>> >>>>>>>>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result
>> in
>> >>>>>>>>>>>>>>>>> the sorting
>> >>>>>>>>>>>>>>>>> in batch case - but - this extension would have to
>> ensure
>> >>>>>>>>>>>>>>>>> the sorting in
>> >>>>>>>>>>>>>>>>> streaming mode also - it would require definition of
>> allowed
>> >>>>>>>>>>>>>>>>> lateness,
>> >>>>>>>>>>>>>>>>> and triggger (essentially similar to window)
>> >>>>>>>>>>>>>>>> This might be reasonable, implemented by default by
>> buffering
>> >>>>>>>>>>>>>>>> everything and releasing elements as the watermark
>> (+lateness)
>> >>>>>>>>>>>>>>>> advances, but would likely lead to inefficient (though
>> >>>>>>>>>>>>>>>> *maybe* easier
>> >>>>>>>>>>>>>>>> to reason about) code. Not sure about the semantics of
>> >>>>>>>>>>>>>>>> triggering
>> >>>>>>>>>>>>>>>> here, especially data-driven triggers. Would it be
>> roughly
>> >>>>>>>>>>>>>>>> equivalent
>> >>>>>>>>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
>> >>>>>>>>>>>>>>>> value in
>> >>>>>>>>>>>>>>>> values])?
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>> Or is the underlying desire just to be able to hint to
>> the
>> >>>>>>>>>>>>>>>> runner that
>> >>>>>>>>>>>>>>>> the code may perform better (e.g. require less
>> resources) as
>> >>>>>>>>>>>>>>>> skew is
>> >>>>>>>>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> - last option would be to introduce these "higher
>> order
>> >>>>>>>>>>>>>>>>> guarantees" in
>> >>>>>>>>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to
>> be the
>> >>>>>>>>>>>>>>>>> worst
>> >>>>>>>>>>>>>>>>> option to me
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> I see the first two options quite equally good,
>> although the
>> >>>>>>>>>>>>>>>>> letter one
>> >>>>>>>>>>>>>>>>> is probably more time consuming to implement. But it
>> would
>> >>>>>>>>>>>>>>>>> bring
>> >>>>>>>>>>>>>>>>> additional feature to streaming case as well.
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> Thanks for any thoughts.
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> Jan
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>> >>>>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>> >>>>>>>>>>>>>>>>>>> Hi Reuven,
>> >>>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
>> >>>>>>>>>>>>>>>>>>>> runners.
>> >>>>>>>>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic
>> inside
>> >>>>>>>>>>>>>>>>>>> the state works for absolutely unbounded
>> out-of-orderness
>> >>>>>>>>>>>>>>>>>>> of elements. That basically (practically) can work
>> only
>> >>>>>>>>>>>>>>>>>>> for cases, where the order of input elements doesn't
>> >>>>>>>>>>>>>>>>>>> matter. But, "state" can refer to "state machine",
>> and any
>> >>>>>>>>>>>>>>>>>>> time you have a state machine involved, then the
>> ordering
>> >>>>>>>>>>>>>>>>>>> of elements would matter.
>> >>>>>>>>>>>>>>>>>> No guarantees on order are provided in *either*
>> streaming
>> >>>>>>>>>>>>>>>>>> or batch
>> >>>>>>>>>>>>>>>>>> mode by the model. However, it is the case that in
>> order to
>> >>>>>>>>>>>>>>>>>> make
>> >>>>>>>>>>>>>>>>>> forward progress most streaming runners attempt to
>> limit
>> >>>>>>>>>>>>>>>>>> the amount of
>> >>>>>>>>>>>>>>>>>> out-of-orderedness of elements (in terms of event time
>> vs.
>> >>>>>>>>>>>>>>>>>> processing
>> >>>>>>>>>>>>>>>>>> time) to make forward progress, which in turn could
>> help
>> >>>>>>>>>>>>>>>>>> cap the
>> >>>>>>>>>>>>>>>>>> amount of state that must be held concurrently,
>> whereas a
>> >>>>>>>>>>>>>>>>>> batch runner
>> >>>>>>>>>>>>>>>>>> may not allow any state to be safely discarded until
>> the whole
>> >>>>>>>>>>>>>>>>>> timeline from infinite past to infinite future has been
>> >>>>>>>>>>>>>>>>>> observed.
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
>> >>>>>>>>>>>>>>>>>> batch" in batch mode.
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>> >>>>>>>>>>>>>>>>>> <mx...@apache.org> wrote:
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>>>> batch semantics and streaming semantics
>> differs only
>> >>>>>>>>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger
>> on
>> >>>>>>>>>>>>>>>>>>>> batch and cannot on stream
>> >>>>>>>>>>>>>>>>>>> You can have a GlobalWindow in streaming with a
>> default
>> >>>>>>>>>>>>>>>>>>> trigger. You
>> >>>>>>>>>>>>>>>>>>> could define additional triggers that do early
>> firings.
>> >>>>>>>>>>>>>>>>>>> And you could
>> >>>>>>>>>>>>>>>>>>> even trigger the global window by advancing the
>> watermark
>> >>>>>>>>>>>>>>>>>>> to +inf.
>> >>>>>>>>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window
>> with
>> >>>>>>>>>>>>>>>>>> default
>> >>>>>>>>>>>>>>>>>> trigger on unbounded PCollections in the SDK because
>> this
>> >>>>>>>>>>>>>>>>>> is more
>> >>>>>>>>>>>>>>>>>> likely to be user error than an actual desire to have
>> no
>> >>>>>>>>>>>>>>>>>> output until
>> >>>>>>>>>>>>>>>>>> drain. But it's semantically valid in the model.
>>
>>
Re: Definition of Unified model
Posted by Jan Lukavský <je...@seznam.cz>.
Hi Reuven,
> It also gets awkward with Flatten - the sequence number is no longer
enough, you must also encode which side of the flatten each element came
from.
That is a generic need. Even if you read data from Kafka, the offsets
are comparable only inside single partition. So, for Kafka to work as a
FIFO for ordering, elements with same key have to be pushed to the same
partition (otherwise Kafka cannot act as FIFO, because different
partitions can be handled by different brokers, which means different
observers and they therefore might not agree on the order of events). So
if we want to emulate FIFO per key, then the sequence IDs have also be
per key.
On 5/28/19 2:33 PM, Reuven Lax wrote:
> Sequence metadata does have the disadvantage that users can no longer
> use the types coming from the source. You must create a new type that
> contains a sequence number (unless Beam provides this). It also gets
> awkward with Flatten - the sequence number is no longer enough, you
> must also encode which side of the flatten each element came from.
>
> On Tue, May 28, 2019 at 3:18 AM Jan Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
>
> As I understood it, Kenn was supporting the idea that sequence
> metadata
> is preferable over FIFO. I was trying to point out, that it even
> should
> provide the same functionally as FIFO, plus one important more -
> reproducibility and ability to being persisted and reused the same
> way
> in batch and streaming.
>
> There is no doubt, that sequence metadata can be stored in every
> storage. But, regarding some implicit ordering that sources might
> have -
> yes, of course, data written into HDFS or Cloud Storage has ordering,
> but only partial - inside some bulk (e.g. file) and the ordering
> is not
> defined correctly on boundaries of these bulks (between files).
> That is
> why I'd say, that ordering of sources is relevant only for
> (partitioned!) streaming sources and generally always reduces to
> sequence metadata (e.g. offsets).
>
> Jan
>
> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
> > Huge +1 to all Kenn said.
> >
> > Jan, batch sources can have orderings too, just like Kafka. I think
> > it's reasonable (for both batch and streaming) that if a source
> has an
> > ordering that is an important part of the data, it should preserve
> > this ordering into the data itself (e.g. as sequence numbers,
> offsets,
> > etc.)
> >
> > On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles
> <kenn@apache.org <ma...@apache.org>> wrote:
> >> I strongly prefer explicit sequence metadata over FIFO
> requirements, because:
> >>
> >> - FIFO is complex to specify: for example Dataflow has "per
> stage key-to-key" FIFO today, but it is not guaranteed to remain
> so (plus "stage" is not a portable concept, nor even guaranteed to
> remain a Dataflow concept)
> >> - complex specifications are by definition poor usability (if
> necessary, then it is what it is)
> >> - overly restricts the runner, reduces parallelism, for
> example any non-stateful ParDo has per-element parallelism, not
> per "key"
> >> - another perspective on that: FIFO makes everyone pay rather
> than just the transform that requires exactly sequencing
> >> - previous implementation details like reshuffles become part
> of the model
> >> - I'm not even convinced the use cases involved are addressed
> by some careful FIFO restrictions; many sinks re-key and they
> would all have to become aware of how keying of a sequence of
> "stages" affects the end-to-end FIFO
> >>
> >> A noop becoming a non-noop is essentially the mathematical
> definition of moving from higher-level to lower-level abstraction.
> >>
> >> So this strikes at the core question of what level of
> abstraction Beam aims to represent. Lower-level means there are
> fewer possible implementations and it is more tied to the
> underlying architecture, and anything not near-exact match pays a
> huge penalty. Higher-level means there are more implementations
> possible with different tradeoffs, though they may all pay a minor
> penalty.
> >>
> >> I could be convinced to change my mind, but it needs some
> extensive design, examples, etc. I think it is probably about the
> most consequential design decision in the whole Beam model, around
> the same level as the decision to use ParDo and GBK as the
> primitives IMO.
> >>
> >> Kenn
> >>
> >> On Thu, May 23, 2019 at 10:17 AM Reuven Lax <relax@google.com
> <ma...@google.com>> wrote:
> >>> Not really. I'm suggesting that some variant of FIFO ordering
> is necessary, which requires either runners natively support FIFO
> ordering or transforms adding some extra sequence number to each
> record to sort by.
> >>>
> >>> I still think your proposal is very useful by the way. I'm
> merely pointing out that to solve the state-machine problem we
> probably need something more.
> >>>
> >>> Reuven
> >>>
> >>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
> >>>> Hi,
> >>>> yes. It seems that ordering by user supplied UDF makes sense
> and I will update the design proposal accordingly.
> >>>> Would that solve the issues you mention?
> >>>> Jan
> >>>> ---------- Původní e-mail ----------
> >>>> Od: Reuven Lax <relax@google.com <ma...@google.com>>
> >>>> Komu: dev <dev@beam.apache.org <ma...@beam.apache.org>>
> >>>> Datum: 23. 5. 2019 18:44:38
> >>>> Předmět: Re: Definition of Unified model
> >>>>
> >>>> I'm simply saying that timestamp ordering is insufficient for
> state machines. I wasn't proposing Kafka as a solution - that was
> simply an example of how people solve this problem in other scenarios.
> >>>>
> >>>> BTW another example of ordering: Imagine today that you have
> a triggered Sum aggregation writing out to a key-value sink. In
> theory we provide no ordering, so the sink might write the
> triggered sums in the wrong order, ending up with an incorrect
> value in the sink. In this case you probably want values ordered
> by trigger pane index.
> >>>>
> >>>> Reuven
> >>>>
> >>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
> >>>>
> >>>> Hi Reuven,
> >>>> I share the view point of Robert. I think the isuue you refer
> to is not in reality related to timestamps, but to the fact, that
> ordering of events in time is observer dependent (either caused by
> relativity, or time skew, essentially this has the same
> consequences). And the resolution in fact isn't Kafka, but
> generally an authoritative observer, that tells you "I saw the
> events in this order". And you either have one (and have the
> outcome of his observation persisted in the data - e.g. as offset
> in Kafka partition), then you should be able to use it (maybe that
> suggests afterall that sorting by some user supplied UDF might
> make sense), or do not have it, and then any interpretation of the
> data seems to be equally valid. Although determinism is fine, of
> course.
> >>>> Jan
> >>>> ---------- Původní e-mail ----------
> >>>> Od: Reuven Lax <relax@google.com <ma...@google.com>>
> >>>> Komu: dev <dev@beam.apache.org <ma...@beam.apache.org>>
> >>>> Datum: 23. 5. 2019 17:39:12
> >>>> Předmět: Re: Definition of Unified model
> >>>>
> >>>> So an example would be elements of type "startUserSession"
> and "endUserSession" (website sessions, not Beam sessions).
> Logically you may need to process them in the correct order if you
> have any sort of state-machine logic. However timestamp ordering
> is never guaranteed to match the logical ordering. Not only might
> you have several elements with the same timestamp, but in reality
> time skew across backend servers can cause the events to have
> timestamps in reverse order of the actual causality order.
> >>>>
> >>>> People do solve this problem today though. Publish the events
> to Kafka, making sure that events for the same user end up in the
> same Kafka partition. This ensures that the events appear in the
> Kafka partitions in causality order, even if the timestamp order
> doesn't match. The your Kafka subscriber simply process the
> elements in each partition in order.
> >>>>
> >>>> I think the ability to impose FIFO causality ordering is
> what's needed for any state-machine work. Timestamp ordering has
> advantages (though often I think the advantage is in state), but
> does not solve this problem.
> >>>>
> >>>> Reuven
> >>>>
> >>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw
> <robertwb@google.com <ma...@google.com>> wrote:
> >>>>
> >>>> Good point.
> >>>>
> >>>> The "implementation-specific" way I would do this is
> >>>> window-by-instant, followed by a DoFn that gets all the
> elements with
> >>>> the same timestamp and sorts/acts accordingly, but this
> counts on the
> >>>> runner producing windows in timestamp order (likely?) and
> also the
> >>>> subsequent DoFn getting them in this order (also likely, due to
> >>>> fusion).
> >>>>
> >>>> One could make the argument that, though it does not provide
> >>>> deterministic behavior, getting elements of the same timestamp in
> >>>> different orders should produce equally valid interpretations
> of the
> >>>> data. (After all, due to relatively, timestamps are not
> technically
> >>>> well ordered across space.) I can see how data-dependent
> tiebreakers
> >>>> could be useful, or promises of preservation of order between
> >>>> operations.
> >>>>
> >>>> - Robert
> >>>>
> >>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <relax@google.com
> <ma...@google.com>> wrote:
> >>>>> So Jan's example of state machines is quite a valid use case
> for ordering. However in my experience, timestamp ordering is
> insufficient for state machines. Elements that cause state
> transitions might come in with the exact same timestamp, yet still
> have a necessary ordering. Especially given Beam's decision to
> have milliseconds timestamps this is possible, but even at
> microsecond or nanosecond precision this can happen at scale. To
> handle state machines you usually need some sort of FIFO ordering
> along with an ordered sources, such as Kafka, not timestamp ordering.
> >>>>>
> >>>>> Reuven
> >>>>>
> >>>>> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský
> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
> >>>>>> Hi all,
> >>>>>>
> >>>>>> thanks everyone for this discussion. I think I have
> gathered enough
> >>>>>> feedback to be able to put down a proposition for changes,
> which I will
> >>>>>> do and send to this list for further discussion. There are
> still doubts
> >>>>>> remaining the non-determinism and it's relation to outputs
> stability vs.
> >>>>>> latency. But I will try to clarify all this in the design
> document.
> >>>>>>
> >>>>>> Thanks,
> >>>>>>
> >>>>>> Jan
> >>>>>>
> >>>>>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
> >>>>>>>> Someone from Flink might correct me if I'm wrong, but
> that's my
> >>>>>>>> current understanding.
> >>>>>>> In essence your description of how exactly-once works in
> Flink is
> >>>>>>> correct. The general assumption in Flink is that pipelines
> must be
> >>>>>>> deterministic and thus produce idempotent writes in the
> case of
> >>>>>>> failures. However, that doesn't mean Beam sinks can't
> guarantee a bit
> >>>>>>> more with what Flink has to offer.
> >>>>>>>
> >>>>>>> Luke already mentioned the design discussions for
> @RequiresStableInput
> >>>>>>> which ensures idempotent writes for non-deterministic
> pipelines. This
> >>>>>>> is not part of the model but an optional Beam feature.
> >>>>>>>
> >>>>>>> We recently implemented support for @RequiresStableInput
> in the Flink
> >>>>>>> Runner. Reuven mentioned the Flink checkpoint
> confirmation, which
> >>>>>>> allows us to buffer (and checkpoint) processed data and
> only emit it
> >>>>>>> once a Flink checkpoint has completed.
> >>>>>>>
> >>>>>>> Cheers,
> >>>>>>> Max
> >>>>>>>
> >>>>>>> On 21.05.19 16:49, Jan Lukavský wrote:
> >>>>>>>> Hi,
> >>>>>>>>
> >>>>>>>> > Actually, I think it is a larger (open) question
> whether exactly
> >>>>>>>> once is guaranteed by the model or whether runners are
> allowed to
> >>>>>>>> relax that. I would think, however, that sources correctly
> >>>>>>>> implemented should be idempotent when run atop an exactly
> once
> >>>>>>>> infrastructure such as Flink of Dataflow.
> >>>>>>>>
> >>>>>>>> I would assume, that the model basically inherits
> guarantees of
> >>>>>>>> underlying infrastructure. Because Flink does not work as you
> >>>>>>>> described (atomic commit of inputs, state and outputs),
> but rather a
> >>>>>>>> checkpoint mark is flowing through the DAG much like
> watermark and on
> >>>>>>>> failures operators are restored and data reprocessed, it
> (IMHO)
> >>>>>>>> implies, that you have exactly once everywhere in the DAG
> *but*
> >>>>>>>> sinks. That is because sinks cannot be restored to
> previous state,
> >>>>>>>> instead sinks are supposed to be idempotent in order for
> the exactly
> >>>>>>>> once to really work (or at least be able to commit outputs on
> >>>>>>>> checkpoint in sink). That implies that if you don't have
> sink that is
> >>>>>>>> able to commit outputs atomically on checkpoint, the pipeline
> >>>>>>>> execution should be deterministic upon retries, otherwise
> shadow
> >>>>>>>> writes from failed paths of the pipeline might appear.
> >>>>>>>>
> >>>>>>>> Someone from Flink might correct me if I'm wrong, but
> that's my
> >>>>>>>> current understanding.
> >>>>>>>>
> >>>>>>>> > Sounds like we should make this clearer.
> >>>>>>>>
> >>>>>>>> I meant that you are right that we must not in any
> thoughts we are
> >>>>>>>> having forget that streams are by definition
> out-of-order. That is
> >>>>>>>> property that we cannot change. But - that doesn't limit
> us from
> >>>>>>>> creating operator that presents the data to UDF as if the
> stream was
> >>>>>>>> ideally sorted. It can do that by introducing latency, of
> course.
> >>>>>>>>
> >>>>>>>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
> >>>>>>>>> Reza: One could provide something like this as a utility
> class, but
> >>>>>>>>> one downside is that it is not scale invariant. It
> requires a tuning
> >>>>>>>>> parameter that, if to small, won't mitigate the problem,
> but if to
> >>>>>>>>> big, greatly increases latency. (Possibly one could
> define a dynamic
> >>>>>>>>> session-like window to solve this though...) It also
> might be harder
> >>>>>>>>> for runners that *can* cheaply present stuff in
> timestamp order to
> >>>>>>>>> optimize. (That and, in practice, our annotation-style
> process methods
> >>>>>>>>> don't lend themselves to easy composition.) I think it
> could work in
> >>>>>>>>> specific cases though.
> >>>>>>>>>
> >>>>>>>>> More inline below.
> >>>>>>>>>
> >>>>>>>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský
> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
> >>>>>>>>>> Hi Robert,
> >>>>>>>>>>
> >>>>>>>>>> > Beam has an exactly-once model. If the data was
> consumed, state
> >>>>>>>>>> mutated, and outputs written downstream (these three
> are committed
> >>>>>>>>>> together atomically) it will not be replayed. That does
> not, of
> >>>>>>>>>> course,
> >>>>>>>>>> solve the non-determanism due to ordering (including
> the fact that two
> >>>>>>>>>> operations reading the same PCollection may view
> different ordering).
> >>>>>>>>>>
> >>>>>>>>>> I think what you describe is a property of a runner,
> not of the model,
> >>>>>>>>>> right? I think if I run my pipeline on Flink I will not
> get this
> >>>>>>>>>> atomicity, because although Flink uses also
> exactly-once model if
> >>>>>>>>>> might
> >>>>>>>>>> write outputs multiple times.
> >>>>>>>>> Actually, I think it is a larger (open) question whether
> exactly once
> >>>>>>>>> is guaranteed by the model or whether runners are
> allowed to relax
> >>>>>>>>> that. I would think, however, that sources correctly
> implemented
> >>>>>>>>> should be idempotent when run atop an exactly once
> infrastructure such
> >>>>>>>>> as Flink of Dataflow.
> >>>>>>>>>
> >>>>>>>>>> > 1) Is it correct for a (Stateful)DoFn to assume
> elements are
> >>>>>>>>>> received
> >>>>>>>>>> in a specific order? In the current model, it is not.
> Being able to
> >>>>>>>>>> read, handle, and produced out-of-order data, including
> late data,
> >>>>>>>>>> is a
> >>>>>>>>>> pretty fundamental property of distributed systems.
> >>>>>>>>>>
> >>>>>>>>>> Yes, absolutely. The argument here is not that Stateful
> ParDo should
> >>>>>>>>>> presume to receive elements in any order, but to
> _present_ it as
> >>>>>>>>>> such to
> >>>>>>>>>> the user @ProcessElement function.
> >>>>>>>>> Sounds like we should make this clearer.
> >>>>>>>>>
> >>>>>>>>>> > 2) Given that some operations are easier (or
> possibly only
> >>>>>>>>>> possible)
> >>>>>>>>>> to write when operating on ordered data, and that
> different runners
> >>>>>>>>>> may
> >>>>>>>>>> have (significantly) cheaper ways to provide this
> ordering than can be
> >>>>>>>>>> done by the user themselves, should we elevate this to
> a property of
> >>>>>>>>>> (Stateful?)DoFns that the runner can provide? I think a
> compelling
> >>>>>>>>>> argument can be made here that we should.
> >>>>>>>>>>
> >>>>>>>>>> +1
> >>>>>>>>>>
> >>>>>>>>>> Jan
> >>>>>>>>>>
> >>>>>>>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
> >>>>>>>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský
> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
> >>>>>>>>>>>> > I don't see batch vs. streaming as part of the
> model. One
> >>>>>>>>>>>> can have
> >>>>>>>>>>>> microbatch, or even a runner that alternates between
> different
> >>>>>>>>>>>> modes.
> >>>>>>>>>>>>
> >>>>>>>>>>>> Although I understand motivation of this statement,
> this project
> >>>>>>>>>>>> name is
> >>>>>>>>>>>> "Apache Beam: An advanced unified programming model".
> What does the
> >>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of
> the model?
> >>>>>>>>>>> What I mean is that streaming vs. batch is no longer
> part of the
> >>>>>>>>>>> model
> >>>>>>>>>>> (or ideally API), but pushed down to be a concern of
> the runner
> >>>>>>>>>>> (executor) of the pipeline.
> >>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský
> <je.ik@seznam.cz <ma...@seznam.cz>>
> >>>>>>>>>>> wrote:
> >>>>>>>>>>>> Hi Kenn,
> >>>>>>>>>>>>
> >>>>>>>>>>>> OK, so if we introduce annotation, we can have
> stateful ParDo
> >>>>>>>>>>>> with sorting, that would perfectly resolve my issues.
> I still
> >>>>>>>>>>>> have some doubts, though. Let me explain. The current
> behavior of
> >>>>>>>>>>>> stateful ParDo has the following properties:
> >>>>>>>>>>>>
> >>>>>>>>>>>> a) might fail in batch, although runs fine in
> streaming (that
> >>>>>>>>>>>> is due to the buffering, and unbounded lateness in
> batch, which
> >>>>>>>>>>>> was discussed back and forth in this thread)
> >>>>>>>>>>>>
> >>>>>>>>>>>> b) might be non deterministic (this is because
> the elements
> >>>>>>>>>>>> arrive at somewhat random order, and even if you do
> the operation
> >>>>>>>>>>>> "assign unique ID to elements" this might produce
> different
> >>>>>>>>>>>> results when run multiple times)
> >>>>>>>>>>> PCollections are *explicitly* unordered. Any
> operations that
> >>>>>>>>>>> assume or
> >>>>>>>>>>> depend on a specific ordering for correctness (or
> determinism) must
> >>>>>>>>>>> provide that ordering themselves (i.e. tolerate
> "arbitrary shuffling
> >>>>>>>>>>> of inputs"). As you point out, that may be very
> expensive if you have
> >>>>>>>>>>> very hot keys with very large (unbounded) timestamp skew.
> >>>>>>>>>>>
> >>>>>>>>>>> StatefulDoFns are low-level operations that should be
> used with care;
> >>>>>>>>>>> the simpler windowing model gives determinism in the
> face of
> >>>>>>>>>>> unordered
> >>>>>>>>>>> data (though late data and non-end-of-window
> triggering introduces
> >>>>>>>>>>> some of the non-determanism back in).
> >>>>>>>>>>>
> >>>>>>>>>>>> What worries me most is the property b), because it
> seems to me
> >>>>>>>>>>>> to have serious consequences - not only that if you
> run twice
> >>>>>>>>>>>> batch pipeline you would get different results, but
> even on
> >>>>>>>>>>>> streaming, when pipeline fails and gets restarted from
> >>>>>>>>>>>> checkpoint, produced output might differ from the
> previous run
> >>>>>>>>>>>> and data from the first run might have already been
> persisted
> >>>>>>>>>>>> into sink. That would create somewhat messy outputs.
> >>>>>>>>>>> Beam has an exactly-once model. If the data was
> consumed, state
> >>>>>>>>>>> mutated, and outputs written downstream (these three
> are committed
> >>>>>>>>>>> together atomically) it will not be replayed. That
> does not, of
> >>>>>>>>>>> course, solve the non-determanism due to ordering
> (including the fact
> >>>>>>>>>>> that two operations reading the same PCollection may
> view different
> >>>>>>>>>>> ordering).
> >>>>>>>>>>>
> >>>>>>>>>>>> These two properties makes me think that the current
> >>>>>>>>>>>> implementation is more of a _special case_ than the
> general one.
> >>>>>>>>>>>> The general one would be that your state doesn't have the
> >>>>>>>>>>>> properties to be able to tolerate buffering problems
> and/or
> >>>>>>>>>>>> non-determinism. Which is the case where you need
> sorting in both
> >>>>>>>>>>>> streaming and batch to be part of the model.
> >>>>>>>>>>>>
> >>>>>>>>>>>> Let me point out one more analogy - that is merging vs.
> >>>>>>>>>>>> non-merging windows. The general case (merging
> windows) implies
> >>>>>>>>>>>> sorting by timestamp in both batch case (explicit)
> and streaming
> >>>>>>>>>>>> (buffering). The special case (non-merging windows)
> doesn't rely
> >>>>>>>>>>>> on any timestamp ordering, so the sorting and
> buffering can be
> >>>>>>>>>>>> dropped. The underlying root cause of this is the
> same for both
> >>>>>>>>>>>> stateful ParDo and windowing (essentially, assigning
> window
> >>>>>>>>>>>> labels is a stateful operation when windowing
> function is merging).
> >>>>>>>>>>>>
> >>>>>>>>>>>> The reason for the current behavior of stateful ParDo
> seems to be
> >>>>>>>>>>>> performance, but is it right to abandon correctness
> in favor of
> >>>>>>>>>>>> performance? Wouldn't it be more consistent to have
> the default
> >>>>>>>>>>>> behavior prefer correctness and when you have the
> specific
> >>>>>>>>>>>> conditions of state function having special
> properties, then you
> >>>>>>>>>>>> can annotate your DoFn (with something like
> >>>>>>>>>>>> @TimeOrderingAgnostic), which would yield a better
> performance in
> >>>>>>>>>>>> that case?
> >>>>>>>>>>> There are two separable questions here.
> >>>>>>>>>>>
> >>>>>>>>>>> 1) Is it correct for a (Stateful)DoFn to assume
> elements are received
> >>>>>>>>>>> in a specific order? In the current model, it is not.
> Being able to
> >>>>>>>>>>> read, handle, and produced out-of-order data,
> including late data, is
> >>>>>>>>>>> a pretty fundamental property of distributed systems.
> >>>>>>>>>>>
> >>>>>>>>>>> 2) Given that some operations are easier (or possibly
> only possible)
> >>>>>>>>>>> to write when operating on ordered data, and that
> different runners
> >>>>>>>>>>> may have (significantly) cheaper ways to provide this
> ordering than
> >>>>>>>>>>> can be done by the user themselves, should we elevate
> this to a
> >>>>>>>>>>> property of (Stateful?)DoFns that the runner can
> provide? I think a
> >>>>>>>>>>> compelling argument can be made here that we should.
> >>>>>>>>>>>
> >>>>>>>>>>> - Robert
> >>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
> >>>>>>>>>>>>
> >>>>>>>>>>>> Thanks for the nice small example of a calculation
> that depends
> >>>>>>>>>>>> on order. You are right that many state machines have
> this
> >>>>>>>>>>>> property. I agree w/ you and Luke that it is
> convenient for batch
> >>>>>>>>>>>> processing to sort by event timestamp before running
> a stateful
> >>>>>>>>>>>> ParDo. In streaming you could also implement "sort by
> event
> >>>>>>>>>>>> timestamp" by buffering until you know all earlier
> data will be
> >>>>>>>>>>>> dropped - a slack buffer up to allowed lateness.
> >>>>>>>>>>>>
> >>>>>>>>>>>> I do not think that it is OK to sort in batch and not in
> >>>>>>>>>>>> streaming. Many state machines diverge very rapidly
> when things
> >>>>>>>>>>>> are out of order. So each runner if they see the
> >>>>>>>>>>>> "@OrderByTimestamp" annotation (or whatever) needs to
> deliver
> >>>>>>>>>>>> sorted data (by some mix of buffering and dropping),
> or to reject
> >>>>>>>>>>>> the pipeline as unsupported.
> >>>>>>>>>>>>
> >>>>>>>>>>>> And also want to say that this is not the default
> case - many
> >>>>>>>>>>>> uses of state & timers in ParDo yield different
> results at the
> >>>>>>>>>>>> element level, but the results are equivalent at in
> the big
> >>>>>>>>>>>> picture. Such as the example of "assign a unique
> sequence number
> >>>>>>>>>>>> to each element" or "group into batches" it doesn't
> matter
> >>>>>>>>>>>> exactly what the result is, only that it meets the
> spec. And
> >>>>>>>>>>>> other cases like user funnels are monotonic enough
> that you also
> >>>>>>>>>>>> don't actually need sorting.
> >>>>>>>>>>>>
> >>>>>>>>>>>> Kenn
> >>>>>>>>>>>>
> >>>>>>>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský
> <je.ik@seznam.cz <ma...@seznam.cz>>
> >>>>>>>>>>>> wrote:
> >>>>>>>>>>>>> Yes, the problem will arise probably mostly when you
> have not
> >>>>>>>>>>>>> well distributed keys (or too few keys). I'm really
> not sure if
> >>>>>>>>>>>>> a pure GBK with a trigger can solve this - it might
> help to have
> >>>>>>>>>>>>> data driven trigger. There would still be some
> doubts, though.
> >>>>>>>>>>>>> The main question is still here - people say, that
> sorting by
> >>>>>>>>>>>>> timestamp before stateful ParDo would be
> prohibitively slow, but
> >>>>>>>>>>>>> I don't really see why - the sorting is very
> probably already
> >>>>>>>>>>>>> there. And if not (hash grouping instead of sorted
> grouping),
> >>>>>>>>>>>>> then the sorting would affect only user defined
> StatefulParDos.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> This would suggest that the best way out of this
> would be really
> >>>>>>>>>>>>> to add annotation, so that the author of the
> pipeline can decide.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> If that would be acceptable I think I can try to
> prepare some
> >>>>>>>>>>>>> basic functionality, but I'm not sure, if I would be
> able to
> >>>>>>>>>>>>> cover all runners / sdks.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> It is read all per key and window and not just read
> all (this
> >>>>>>>>>>>>> still won't scale with hot keys in the global
> window). The GBK
> >>>>>>>>>>>>> preceding the StatefulParDo will guarantee that you are
> >>>>>>>>>>>>> processing all the values for a specific key and
> window at any
> >>>>>>>>>>>>> given time. Is there a specific window/trigger that
> is missing
> >>>>>>>>>>>>> that you feel would remove the need for you to use
> StatefulParDo?
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský
> <je.ik@seznam.cz <ma...@seznam.cz>>
> >>>>>>>>>>>>> wrote:
> >>>>>>>>>>>>>> Hi Lukasz,
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Today, if you must have a strict order, you must
> guarantee
> >>>>>>>>>>>>>>> that your StatefulParDo implements the necessary
> "buffering &
> >>>>>>>>>>>>>>> sorting" into state.
> >>>>>>>>>>>>>> Yes, no problem with that. But this whole
> discussion started,
> >>>>>>>>>>>>>> because *this doesn't work on batch*. You simply
> cannot first
> >>>>>>>>>>>>>> read everything from distributed storage and then
> buffer it all
> >>>>>>>>>>>>>> into memory, just to read it again, but sorted.
> That will not
> >>>>>>>>>>>>>> work. And even if it would, it would be a terrible
> waste of
> >>>>>>>>>>>>>> resources.
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> Jan
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský
> <je.ik@seznam.cz <ma...@seznam.cz>>
> >>>>>>>>>>>>>> wrote:
> >>>>>>>>>>>>>>> This discussion brings many really interesting
> questions for
> >>>>>>>>>>>>>>> me. :-)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > I don't see batch vs. streaming as part of
> the model. One
> >>>>>>>>>>>>>>> can have
> >>>>>>>>>>>>>>> microbatch, or even a runner that alternates
> between different
> >>>>>>>>>>>>>>> modes.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Although I understand motivation of this
> statement, this
> >>>>>>>>>>>>>>> project name is
> >>>>>>>>>>>>>>> "Apache Beam: An advanced unified programming
> model". What
> >>>>>>>>>>>>>>> does the
> >>>>>>>>>>>>>>> model unify, if "streaming vs. batch" is not part
> of the model?
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Using microbatching, chaining of batch jobs, or
> pure streaming
> >>>>>>>>>>>>>>> are
> >>>>>>>>>>>>>>> exactly the "runtime conditions/characteristics" I
> refer to.
> >>>>>>>>>>>>>>> All these
> >>>>>>>>>>>>>>> define several runtime parameters, which in turn
> define how
> >>>>>>>>>>>>>>> well/badly
> >>>>>>>>>>>>>>> will the pipeline perform and how many resources
> might be
> >>>>>>>>>>>>>>> needed. From
> >>>>>>>>>>>>>>> my point of view, pure streaming should be the
> most resource
> >>>>>>>>>>>>>>> demanding
> >>>>>>>>>>>>>>> (if not, why bother with batch? why not run
> everything in
> >>>>>>>>>>>>>>> streaming
> >>>>>>>>>>>>>>> only? what will there remain to "unify"?).
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > Fortunately, for batch, only the state for a
> single key
> >>>>>>>>>>>>>>> needs to be
> >>>>>>>>>>>>>>> preserved at a time, rather than the state for all
> keys across
> >>>>>>>>>>>>>>> the range
> >>>>>>>>>>>>>>> of skew. Of course if you have few or hot keys,
> one can still
> >>>>>>>>>>>>>>> have
> >>>>>>>>>>>>>>> issues (and this is not specific to StatefulDoFns).
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Yes, but here is still the presumption that my
> stateful DoFn can
> >>>>>>>>>>>>>>> tolerate arbitrary shuffling of inputs. Let me
> explain the use
> >>>>>>>>>>>>>>> case in
> >>>>>>>>>>>>>>> more detail.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Suppose you have input stream consisting of 1s and
> 0s (and
> >>>>>>>>>>>>>>> some key for
> >>>>>>>>>>>>>>> each element, which is irrelevant for the
> demonstration). Your
> >>>>>>>>>>>>>>> task is
> >>>>>>>>>>>>>>> to calculate in running global window the actual
> number of
> >>>>>>>>>>>>>>> changes
> >>>>>>>>>>>>>>> between state 0 and state 1 and vice versa. When
> the state
> >>>>>>>>>>>>>>> doesn't
> >>>>>>>>>>>>>>> change, you don't calculate anything. If input
> (for given key)
> >>>>>>>>>>>>>>> would be
> >>>>>>>>>>>>>>> (tN denotes timestamp N):
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t1: 1
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t2: 0
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t3: 0
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t4: 1
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t5: 1
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t6: 0
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> then the output should yield (supposing that
> default state is
> >>>>>>>>>>>>>>> zero):
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t1: (one: 1, zero: 0)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t2: (one: 1, zero: 1)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t3: (one: 1, zero: 1)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t4: (one: 2, zero: 1)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t5: (one: 2, zero: 1)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t6: (one: 2, zero: 2)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> How would you implement this in current Beam
> semantics?
> >>>>>>>>>>>>>> I think your saying here that I know that my input
> is ordered
> >>>>>>>>>>>>>> in a specific way and since I assume the order when
> writing my
> >>>>>>>>>>>>>> pipeline I can perform this optimization. But there
> is nothing
> >>>>>>>>>>>>>> preventing a runner from noticing that your
> processing in the
> >>>>>>>>>>>>>> global window with a specific type of trigger and
> re-ordering
> >>>>>>>>>>>>>> your inputs/processing to get better performance
> (since you
> >>>>>>>>>>>>>> can't use an AfterWatermark trigger for your
> pipeline in
> >>>>>>>>>>>>>> streaming for the GlobalWindow).
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> Today, if you must have a strict order, you must
> guarantee that
> >>>>>>>>>>>>>> your StatefulParDo implements the necessary
> "buffering &
> >>>>>>>>>>>>>> sorting" into state. I can see why you would want
> an annotation
> >>>>>>>>>>>>>> that says I must have timestamp ordered elements,
> since it
> >>>>>>>>>>>>>> makes writing certain StatefulParDos much easier.
> StatefulParDo
> >>>>>>>>>>>>>> is a low-level function, it really is the "here you
> go and do
> >>>>>>>>>>>>>> whatever you need to but here be dragons" function
> while
> >>>>>>>>>>>>>> windowing and triggering is meant to keep many
> people from
> >>>>>>>>>>>>>> writing StatefulParDo in the first place.
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > Pipelines that fail in the "worst case" batch
> scenario
> >>>>>>>>>>>>>>> are likely to
> >>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when
> the watermark
> >>>>>>>>>>>>>>> falls
> >>>>>>>>>>>>>>> behind in streaming mode as well.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> But the worst case is defined by input of size
> (available
> >>>>>>>>>>>>>>> resources +
> >>>>>>>>>>>>>>> single byte) -> pipeline fail. Although it could have
> >>>>>>>>>>>>>>> finished, given
> >>>>>>>>>>>>>>> the right conditions.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > This might be reasonable, implemented by
> default by
> >>>>>>>>>>>>>>> buffering
> >>>>>>>>>>>>>>> everything and releasing elements as the watermark
> (+lateness)
> >>>>>>>>>>>>>>> advances,
> >>>>>>>>>>>>>>> but would likely lead to inefficient (though
> *maybe* easier to
> >>>>>>>>>>>>>>> reason
> >>>>>>>>>>>>>>> about) code.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Sure, the pipeline will be less efficient, because
> it would
> >>>>>>>>>>>>>>> have to
> >>>>>>>>>>>>>>> buffer and sort the inputs. But at least it will
> produce
> >>>>>>>>>>>>>>> correct results
> >>>>>>>>>>>>>>> in cases where updates to state are order-sensitive.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > Would it be roughly equivalent to GBK +
> FlatMap(lambda
> >>>>>>>>>>>>>>> (key, values):
> >>>>>>>>>>>>>>> [(key, value) for value in values])?
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> I'd say roughly yes, but difference would be in
> the trigger.
> >>>>>>>>>>>>>>> The trigger
> >>>>>>>>>>>>>>> should ideally fire as soon as watermark
> (+lateness) crosses
> >>>>>>>>>>>>>>> element
> >>>>>>>>>>>>>>> with lowest timestamp in the buffer. Although this
> could be
> >>>>>>>>>>>>>>> somehow
> >>>>>>>>>>>>>>> emulated by fixed trigger each X millis.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > Or is the underlying desire just to be able
> to hint to
> >>>>>>>>>>>>>>> the runner
> >>>>>>>>>>>>>>> that the code may perform better (e.g. require
> less resources)
> >>>>>>>>>>>>>>> as skew
> >>>>>>>>>>>>>>> is reduced (and hence to order by timestamp iff
> it's cheap)?
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> No, the sorting would have to be done in streaming
> case as
> >>>>>>>>>>>>>>> well. That is
> >>>>>>>>>>>>>>> an imperative of the unified model. I think it is
> possible to
> >>>>>>>>>>>>>>> sort by
> >>>>>>>>>>>>>>> timestamp only in batch case (and do it for *all*
> batch
> >>>>>>>>>>>>>>> stateful pardos
> >>>>>>>>>>>>>>> without annotation), or introduce annotation, but
> then make
> >>>>>>>>>>>>>>> the same
> >>>>>>>>>>>>>>> guarantees for streaming case as well.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Jan
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
> >>>>>>>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
> >>>>>>>>>>>>>>>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
> >>>>>>>>>>>>>>>>> Hi Robert,
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> yes, I think you rephrased my point - although
> no *explicit*
> >>>>>>>>>>>>>>>>> guarantees
> >>>>>>>>>>>>>>>>> of ordering are given in either mode, there is
> *implicit*
> >>>>>>>>>>>>>>>>> ordering in
> >>>>>>>>>>>>>>>>> streaming case that is due to nature of the
> processing - the
> >>>>>>>>>>>>>>>>> difference
> >>>>>>>>>>>>>>>>> between watermark and timestamp of elements
> flowing through
> >>>>>>>>>>>>>>>>> the pipeline
> >>>>>>>>>>>>>>>>> are generally low (too high difference leads to the
> >>>>>>>>>>>>>>>>> overbuffering
> >>>>>>>>>>>>>>>>> problem), but there is no such bound on batch.
> >>>>>>>>>>>>>>>> Fortunately, for batch, only the state for a
> single key needs
> >>>>>>>>>>>>>>>> to be
> >>>>>>>>>>>>>>>> preserved at a time, rather than the state for
> all keys
> >>>>>>>>>>>>>>>> across the
> >>>>>>>>>>>>>>>> range of skew. Of course if you have few or hot
> keys, one can
> >>>>>>>>>>>>>>>> still
> >>>>>>>>>>>>>>>> have issues (and this is not specific to
> StatefulDoFns).
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> As a result, I see a few possible solutions:
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> - the best and most natural seems to be
> extension of
> >>>>>>>>>>>>>>>>> the model, so
> >>>>>>>>>>>>>>>>> that it defines batch as not only "streaming
> pipeline
> >>>>>>>>>>>>>>>>> executed in batch
> >>>>>>>>>>>>>>>>> fashion", but "pipeline with at least as good
> runtime
> >>>>>>>>>>>>>>>>> characteristics as
> >>>>>>>>>>>>>>>>> in streaming case, executed in batch fashion", I
> really
> >>>>>>>>>>>>>>>>> don't think that
> >>>>>>>>>>>>>>>>> there are any conflicts with the current model,
> or that this
> >>>>>>>>>>>>>>>>> could
> >>>>>>>>>>>>>>>>> affect performance, because the required sorting
> (as pointed by
> >>>>>>>>>>>>>>>>> Aljoscha) is very probably already done during
> translation
> >>>>>>>>>>>>>>>>> of stateful
> >>>>>>>>>>>>>>>>> pardos. Also note that this definition only
> affects user
> >>>>>>>>>>>>>>>>> defined
> >>>>>>>>>>>>>>>>> stateful pardos
> >>>>>>>>>>>>>>>> I don't see batch vs. streaming as part of the
> model. One can
> >>>>>>>>>>>>>>>> have
> >>>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
> >>>>>>>>>>>>>>>> different modes.
> >>>>>>>>>>>>>>>> The model describes what the valid outputs are
> given a
> >>>>>>>>>>>>>>>> (sometimes
> >>>>>>>>>>>>>>>> partial) set of inputs. It becomes really hard to
> define
> >>>>>>>>>>>>>>>> things like
> >>>>>>>>>>>>>>>> "as good runtime characteristics." Once you allow any
> >>>>>>>>>>>>>>>> out-of-orderedness, it is not very feasible to
> try and define
> >>>>>>>>>>>>>>>> (and
> >>>>>>>>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
> >>>>>>>>>>>>>>>> out-of-orderedness.
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> Pipelines that fail in the "worst case" batch
> scenario are
> >>>>>>>>>>>>>>>> likely to
> >>>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when
> the watermark
> >>>>>>>>>>>>>>>> falls
> >>>>>>>>>>>>>>>> behind in streaming mode as well.
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> - another option would be to introduce
> annotation for
> >>>>>>>>>>>>>>>>> DoFns (e.g.
> >>>>>>>>>>>>>>>>> @RequiresStableTimeCharacteristics), which would
> result in
> >>>>>>>>>>>>>>>>> the sorting
> >>>>>>>>>>>>>>>>> in batch case - but - this extension would have
> to ensure
> >>>>>>>>>>>>>>>>> the sorting in
> >>>>>>>>>>>>>>>>> streaming mode also - it would require
> definition of allowed
> >>>>>>>>>>>>>>>>> lateness,
> >>>>>>>>>>>>>>>>> and triggger (essentially similar to window)
> >>>>>>>>>>>>>>>> This might be reasonable, implemented by default
> by buffering
> >>>>>>>>>>>>>>>> everything and releasing elements as the
> watermark (+lateness)
> >>>>>>>>>>>>>>>> advances, but would likely lead to inefficient
> (though
> >>>>>>>>>>>>>>>> *maybe* easier
> >>>>>>>>>>>>>>>> to reason about) code. Not sure about the
> semantics of
> >>>>>>>>>>>>>>>> triggering
> >>>>>>>>>>>>>>>> here, especially data-driven triggers. Would it
> be roughly
> >>>>>>>>>>>>>>>> equivalent
> >>>>>>>>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key,
> value) for
> >>>>>>>>>>>>>>>> value in
> >>>>>>>>>>>>>>>> values])?
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> Or is the underlying desire just to be able to
> hint to the
> >>>>>>>>>>>>>>>> runner that
> >>>>>>>>>>>>>>>> the code may perform better (e.g. require less
> resources) as
> >>>>>>>>>>>>>>>> skew is
> >>>>>>>>>>>>>>>> reduced (and hence to order by timestamp iff it's
> cheap)?
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> - last option would be to introduce these
> "higher order
> >>>>>>>>>>>>>>>>> guarantees" in
> >>>>>>>>>>>>>>>>> some extension DSL (e.g. Euphoria), but that
> seems to be the
> >>>>>>>>>>>>>>>>> worst
> >>>>>>>>>>>>>>>>> option to me
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> I see the first two options quite equally good,
> although the
> >>>>>>>>>>>>>>>>> letter one
> >>>>>>>>>>>>>>>>> is probably more time consuming to implement.
> But it would
> >>>>>>>>>>>>>>>>> bring
> >>>>>>>>>>>>>>>>> additional feature to streaming case as well.
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> Thanks for any thoughts.
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> Jan
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
> >>>>>>>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
> >>>>>>>>>>>>>>>>>> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
> >>>>>>>>>>>>>>>>>>> Hi Reuven,
> >>>>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine
> in batch
> >>>>>>>>>>>>>>>>>>>> runners.
> >>>>>>>>>>>>>>>>>>> Stateful ParDo works in batch as far, as the
> logic inside
> >>>>>>>>>>>>>>>>>>> the state works for absolutely unbounded
> out-of-orderness
> >>>>>>>>>>>>>>>>>>> of elements. That basically (practically) can
> work only
> >>>>>>>>>>>>>>>>>>> for cases, where the order of input elements
> doesn't
> >>>>>>>>>>>>>>>>>>> matter. But, "state" can refer to "state
> machine", and any
> >>>>>>>>>>>>>>>>>>> time you have a state machine involved, then
> the ordering
> >>>>>>>>>>>>>>>>>>> of elements would matter.
> >>>>>>>>>>>>>>>>>> No guarantees on order are provided in *either*
> streaming
> >>>>>>>>>>>>>>>>>> or batch
> >>>>>>>>>>>>>>>>>> mode by the model. However, it is the case that
> in order to
> >>>>>>>>>>>>>>>>>> make
> >>>>>>>>>>>>>>>>>> forward progress most streaming runners attempt
> to limit
> >>>>>>>>>>>>>>>>>> the amount of
> >>>>>>>>>>>>>>>>>> out-of-orderedness of elements (in terms of
> event time vs.
> >>>>>>>>>>>>>>>>>> processing
> >>>>>>>>>>>>>>>>>> time) to make forward progress, which in turn
> could help
> >>>>>>>>>>>>>>>>>> cap the
> >>>>>>>>>>>>>>>>>> amount of state that must be held concurrently,
> whereas a
> >>>>>>>>>>>>>>>>>> batch runner
> >>>>>>>>>>>>>>>>>> may not allow any state to be safely discarded
> until the whole
> >>>>>>>>>>>>>>>>>> timeline from infinite past to infinite future
> has been
> >>>>>>>>>>>>>>>>>> observed.
> >>>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>>> Also, as pointed out, state is not preserved
> "batch to
> >>>>>>>>>>>>>>>>>> batch" in batch mode.
> >>>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
> >>>>>>>>>>>>>>>>>> <mxm@apache.org <ma...@apache.org>> wrote:
> >>>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>>>>> batch semantics and streaming semantics
> differs only
> >>>>>>>>>>>>>>>>>>>> in that I can have GlobalWindow with default
> trigger on
> >>>>>>>>>>>>>>>>>>>> batch and cannot on stream
> >>>>>>>>>>>>>>>>>>> You can have a GlobalWindow in streaming with
> a default
> >>>>>>>>>>>>>>>>>>> trigger. You
> >>>>>>>>>>>>>>>>>>> could define additional triggers that do early
> firings.
> >>>>>>>>>>>>>>>>>>> And you could
> >>>>>>>>>>>>>>>>>>> even trigger the global window by advancing
> the watermark
> >>>>>>>>>>>>>>>>>>> to +inf.
> >>>>>>>>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global
> window with
> >>>>>>>>>>>>>>>>>> default
> >>>>>>>>>>>>>>>>>> trigger on unbounded PCollections in the SDK
> because this
> >>>>>>>>>>>>>>>>>> is more
> >>>>>>>>>>>>>>>>>> likely to be user error than an actual desire
> to have no
> >>>>>>>>>>>>>>>>>> output until
> >>>>>>>>>>>>>>>>>> drain. But it's semantically valid in the model.
>
Re: Definition of Unified model
Posted by Robert Bradshaw <ro...@google.com>.
On Tue, May 28, 2019 at 12:18 PM Jan Lukavský <je...@seznam.cz> wrote:
>
> As I understood it, Kenn was supporting the idea that sequence metadata
> is preferable over FIFO. I was trying to point out, that it even should
> provide the same functionally as FIFO, plus one important more -
> reproducibility and ability to being persisted and reused the same way
> in batch and streaming.
>
> There is no doubt, that sequence metadata can be stored in every
> storage. But, regarding some implicit ordering that sources might have -
> yes, of course, data written into HDFS or Cloud Storage has ordering,
> but only partial - inside some bulk (e.g. file) and the ordering is not
> defined correctly on boundaries of these bulks (between files). That is
> why I'd say, that ordering of sources is relevant only for
> (partitioned!) streaming sources and generally always reduces to
> sequence metadata (e.g. offsets).
Offsets within a file, unordered between files seems exactly analogous
with offsets within a partition, unordered between partitions, right?
> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
> > Huge +1 to all Kenn said.
> >
> > Jan, batch sources can have orderings too, just like Kafka. I think
> > it's reasonable (for both batch and streaming) that if a source has an
> > ordering that is an important part of the data, it should preserve
> > this ordering into the data itself (e.g. as sequence numbers, offsets,
> > etc.)
> >
> > On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org> wrote:
> >> I strongly prefer explicit sequence metadata over FIFO requirements, because:
> >>
> >> - FIFO is complex to specify: for example Dataflow has "per stage key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage" is not a portable concept, nor even guaranteed to remain a Dataflow concept)
> >> - complex specifications are by definition poor usability (if necessary, then it is what it is)
> >> - overly restricts the runner, reduces parallelism, for example any non-stateful ParDo has per-element parallelism, not per "key"
> >> - another perspective on that: FIFO makes everyone pay rather than just the transform that requires exactly sequencing
> >> - previous implementation details like reshuffles become part of the model
> >> - I'm not even convinced the use cases involved are addressed by some careful FIFO restrictions; many sinks re-key and they would all have to become aware of how keying of a sequence of "stages" affects the end-to-end FIFO
> >>
> >> A noop becoming a non-noop is essentially the mathematical definition of moving from higher-level to lower-level abstraction.
> >>
> >> So this strikes at the core question of what level of abstraction Beam aims to represent. Lower-level means there are fewer possible implementations and it is more tied to the underlying architecture, and anything not near-exact match pays a huge penalty. Higher-level means there are more implementations possible with different tradeoffs, though they may all pay a minor penalty.
> >>
> >> I could be convinced to change my mind, but it needs some extensive design, examples, etc. I think it is probably about the most consequential design decision in the whole Beam model, around the same level as the decision to use ParDo and GBK as the primitives IMO.
> >>
> >> Kenn
Re: Definition of Unified model
Posted by Robert Bradshaw <ro...@google.com>.
On Tue, May 28, 2019 at 2:34 PM Reuven Lax <re...@google.com> wrote:
>
> Sequence metadata does have the disadvantage that users can no longer use the types coming from the source. You must create a new type that contains a sequence number (unless Beam provides this).
Yes. Well, the source would have a withSequenceNumber or withOffest or
similar to provide this capability.
> It also gets awkward with Flatten - the sequence number is no
longer enough, you must also encode which side of the flatten each
element came from.
Only if you want to order things across different PCollections. But,
yes, you'd probably want to do a tagged flatten.
I still think making it explicit, and providing the ability to
(cheaply) order on it, is better than trying to come up with an
implicit model of where things are and aren't ordered.
(As an aside, we often have various pieces of metadata that are
attached to elements: timestamps, windows, pane infos, possible
trigger metadata (for data-driven triggers), sequence numbers, ...).
It may be interesting to have the ability to project an element into a
"visible" portion and an invisible "attached" portion, where the
latter would flow through just as timestamps do (where possible), and
then later make it visible again. Maybe the work with schema field
unpacking, etc. to only look at the fields one is interested in is
sufficient to sideline the noise of extra fields, but it wouldn't
allow the application of a (more) arbitrary PTransform<T> to a
PCollection<T + hidden extra data>.)
> On Tue, May 28, 2019 at 3:18 AM Jan Lukavský <je...@seznam.cz> wrote:
>>
>> As I understood it, Kenn was supporting the idea that sequence metadata
>> is preferable over FIFO. I was trying to point out, that it even should
>> provide the same functionally as FIFO, plus one important more -
>> reproducibility and ability to being persisted and reused the same way
>> in batch and streaming.
>>
>> There is no doubt, that sequence metadata can be stored in every
>> storage. But, regarding some implicit ordering that sources might have -
>> yes, of course, data written into HDFS or Cloud Storage has ordering,
>> but only partial - inside some bulk (e.g. file) and the ordering is not
>> defined correctly on boundaries of these bulks (between files). That is
>> why I'd say, that ordering of sources is relevant only for
>> (partitioned!) streaming sources and generally always reduces to
>> sequence metadata (e.g. offsets).
>>
>> Jan
>>
>> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
>> > Huge +1 to all Kenn said.
>> >
>> > Jan, batch sources can have orderings too, just like Kafka. I think
>> > it's reasonable (for both batch and streaming) that if a source has an
>> > ordering that is an important part of the data, it should preserve
>> > this ordering into the data itself (e.g. as sequence numbers, offsets,
>> > etc.)
>> >
>> > On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org> wrote:
>> >> I strongly prefer explicit sequence metadata over FIFO requirements, because:
>> >>
>> >> - FIFO is complex to specify: for example Dataflow has "per stage key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage" is not a portable concept, nor even guaranteed to remain a Dataflow concept)
>> >> - complex specifications are by definition poor usability (if necessary, then it is what it is)
>> >> - overly restricts the runner, reduces parallelism, for example any non-stateful ParDo has per-element parallelism, not per "key"
>> >> - another perspective on that: FIFO makes everyone pay rather than just the transform that requires exactly sequencing
>> >> - previous implementation details like reshuffles become part of the model
>> >> - I'm not even convinced the use cases involved are addressed by some careful FIFO restrictions; many sinks re-key and they would all have to become aware of how keying of a sequence of "stages" affects the end-to-end FIFO
>> >>
>> >> A noop becoming a non-noop is essentially the mathematical definition of moving from higher-level to lower-level abstraction.
>> >>
>> >> So this strikes at the core question of what level of abstraction Beam aims to represent. Lower-level means there are fewer possible implementations and it is more tied to the underlying architecture, and anything not near-exact match pays a huge penalty. Higher-level means there are more implementations possible with different tradeoffs, though they may all pay a minor penalty.
>> >>
>> >> I could be convinced to change my mind, but it needs some extensive design, examples, etc. I think it is probably about the most consequential design decision in the whole Beam model, around the same level as the decision to use ParDo and GBK as the primitives IMO.
>> >>
>> >> Kenn
>> >>
>> >> On Thu, May 23, 2019 at 10:17 AM Reuven Lax <re...@google.com> wrote:
>> >>> Not really. I'm suggesting that some variant of FIFO ordering is necessary, which requires either runners natively support FIFO ordering or transforms adding some extra sequence number to each record to sort by.
>> >>>
>> >>> I still think your proposal is very useful by the way. I'm merely pointing out that to solve the state-machine problem we probably need something more.
>> >>>
>> >>> Reuven
>> >>>
>> >>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je...@seznam.cz> wrote:
>> >>>> Hi,
>> >>>> yes. It seems that ordering by user supplied UDF makes sense and I will update the design proposal accordingly.
>> >>>> Would that solve the issues you mention?
>> >>>> Jan
>> >>>> ---------- Původní e-mail ----------
>> >>>> Od: Reuven Lax <re...@google.com>
>> >>>> Komu: dev <de...@beam.apache.org>
>> >>>> Datum: 23. 5. 2019 18:44:38
>> >>>> Předmět: Re: Definition of Unified model
>> >>>>
>> >>>> I'm simply saying that timestamp ordering is insufficient for state machines. I wasn't proposing Kafka as a solution - that was simply an example of how people solve this problem in other scenarios.
>> >>>>
>> >>>> BTW another example of ordering: Imagine today that you have a triggered Sum aggregation writing out to a key-value sink. In theory we provide no ordering, so the sink might write the triggered sums in the wrong order, ending up with an incorrect value in the sink. In this case you probably want values ordered by trigger pane index.
>> >>>>
>> >>>> Reuven
>> >>>>
>> >>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz> wrote:
>> >>>>
>> >>>> Hi Reuven,
>> >>>> I share the view point of Robert. I think the isuue you refer to is not in reality related to timestamps, but to the fact, that ordering of events in time is observer dependent (either caused by relativity, or time skew, essentially this has the same consequences). And the resolution in fact isn't Kafka, but generally an authoritative observer, that tells you "I saw the events in this order". And you either have one (and have the outcome of his observation persisted in the data - e.g. as offset in Kafka partition), then you should be able to use it (maybe that suggests afterall that sorting by some user supplied UDF might make sense), or do not have it, and then any interpretation of the data seems to be equally valid. Although determinism is fine, of course.
>> >>>> Jan
>> >>>> ---------- Původní e-mail ----------
>> >>>> Od: Reuven Lax <re...@google.com>
>> >>>> Komu: dev <de...@beam.apache.org>
>> >>>> Datum: 23. 5. 2019 17:39:12
>> >>>> Předmět: Re: Definition of Unified model
>> >>>>
>> >>>> So an example would be elements of type "startUserSession" and "endUserSession" (website sessions, not Beam sessions). Logically you may need to process them in the correct order if you have any sort of state-machine logic. However timestamp ordering is never guaranteed to match the logical ordering. Not only might you have several elements with the same timestamp, but in reality time skew across backend servers can cause the events to have timestamps in reverse order of the actual causality order.
>> >>>>
>> >>>> People do solve this problem today though. Publish the events to Kafka, making sure that events for the same user end up in the same Kafka partition. This ensures that the events appear in the Kafka partitions in causality order, even if the timestamp order doesn't match. The your Kafka subscriber simply process the elements in each partition in order.
>> >>>>
>> >>>> I think the ability to impose FIFO causality ordering is what's needed for any state-machine work. Timestamp ordering has advantages (though often I think the advantage is in state), but does not solve this problem.
>> >>>>
>> >>>> Reuven
>> >>>>
>> >>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <ro...@google.com> wrote:
>> >>>>
>> >>>> Good point.
>> >>>>
>> >>>> The "implementation-specific" way I would do this is
>> >>>> window-by-instant, followed by a DoFn that gets all the elements with
>> >>>> the same timestamp and sorts/acts accordingly, but this counts on the
>> >>>> runner producing windows in timestamp order (likely?) and also the
>> >>>> subsequent DoFn getting them in this order (also likely, due to
>> >>>> fusion).
>> >>>>
>> >>>> One could make the argument that, though it does not provide
>> >>>> deterministic behavior, getting elements of the same timestamp in
>> >>>> different orders should produce equally valid interpretations of the
>> >>>> data. (After all, due to relatively, timestamps are not technically
>> >>>> well ordered across space.) I can see how data-dependent tiebreakers
>> >>>> could be useful, or promises of preservation of order between
>> >>>> operations.
>> >>>>
>> >>>> - Robert
>> >>>>
>> >>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com> wrote:
>> >>>>> So Jan's example of state machines is quite a valid use case for ordering. However in my experience, timestamp ordering is insufficient for state machines. Elements that cause state transitions might come in with the exact same timestamp, yet still have a necessary ordering. Especially given Beam's decision to have milliseconds timestamps this is possible, but even at microsecond or nanosecond precision this can happen at scale. To handle state machines you usually need some sort of FIFO ordering along with an ordered sources, such as Kafka, not timestamp ordering.
>> >>>>>
>> >>>>> Reuven
>> >>>>>
>> >>>>> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz> wrote:
>> >>>>>> Hi all,
>> >>>>>>
>> >>>>>> thanks everyone for this discussion. I think I have gathered enough
>> >>>>>> feedback to be able to put down a proposition for changes, which I will
>> >>>>>> do and send to this list for further discussion. There are still doubts
>> >>>>>> remaining the non-determinism and it's relation to outputs stability vs.
>> >>>>>> latency. But I will try to clarify all this in the design document.
>> >>>>>>
>> >>>>>> Thanks,
>> >>>>>>
>> >>>>>> Jan
>> >>>>>>
>> >>>>>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
>> >>>>>>>> current understanding.
>> >>>>>>> In essence your description of how exactly-once works in Flink is
>> >>>>>>> correct. The general assumption in Flink is that pipelines must be
>> >>>>>>> deterministic and thus produce idempotent writes in the case of
>> >>>>>>> failures. However, that doesn't mean Beam sinks can't guarantee a bit
>> >>>>>>> more with what Flink has to offer.
>> >>>>>>>
>> >>>>>>> Luke already mentioned the design discussions for @RequiresStableInput
>> >>>>>>> which ensures idempotent writes for non-deterministic pipelines. This
>> >>>>>>> is not part of the model but an optional Beam feature.
>> >>>>>>>
>> >>>>>>> We recently implemented support for @RequiresStableInput in the Flink
>> >>>>>>> Runner. Reuven mentioned the Flink checkpoint confirmation, which
>> >>>>>>> allows us to buffer (and checkpoint) processed data and only emit it
>> >>>>>>> once a Flink checkpoint has completed.
>> >>>>>>>
>> >>>>>>> Cheers,
>> >>>>>>> Max
>> >>>>>>>
>> >>>>>>> On 21.05.19 16:49, Jan Lukavský wrote:
>> >>>>>>>> Hi,
>> >>>>>>>>
>> >>>>>>>> > Actually, I think it is a larger (open) question whether exactly
>> >>>>>>>> once is guaranteed by the model or whether runners are allowed to
>> >>>>>>>> relax that. I would think, however, that sources correctly
>> >>>>>>>> implemented should be idempotent when run atop an exactly once
>> >>>>>>>> infrastructure such as Flink of Dataflow.
>> >>>>>>>>
>> >>>>>>>> I would assume, that the model basically inherits guarantees of
>> >>>>>>>> underlying infrastructure. Because Flink does not work as you
>> >>>>>>>> described (atomic commit of inputs, state and outputs), but rather a
>> >>>>>>>> checkpoint mark is flowing through the DAG much like watermark and on
>> >>>>>>>> failures operators are restored and data reprocessed, it (IMHO)
>> >>>>>>>> implies, that you have exactly once everywhere in the DAG *but*
>> >>>>>>>> sinks. That is because sinks cannot be restored to previous state,
>> >>>>>>>> instead sinks are supposed to be idempotent in order for the exactly
>> >>>>>>>> once to really work (or at least be able to commit outputs on
>> >>>>>>>> checkpoint in sink). That implies that if you don't have sink that is
>> >>>>>>>> able to commit outputs atomically on checkpoint, the pipeline
>> >>>>>>>> execution should be deterministic upon retries, otherwise shadow
>> >>>>>>>> writes from failed paths of the pipeline might appear.
>> >>>>>>>>
>> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
>> >>>>>>>> current understanding.
>> >>>>>>>>
>> >>>>>>>> > Sounds like we should make this clearer.
>> >>>>>>>>
>> >>>>>>>> I meant that you are right that we must not in any thoughts we are
>> >>>>>>>> having forget that streams are by definition out-of-order. That is
>> >>>>>>>> property that we cannot change. But - that doesn't limit us from
>> >>>>>>>> creating operator that presents the data to UDF as if the stream was
>> >>>>>>>> ideally sorted. It can do that by introducing latency, of course.
>> >>>>>>>>
>> >>>>>>>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>> >>>>>>>>> Reza: One could provide something like this as a utility class, but
>> >>>>>>>>> one downside is that it is not scale invariant. It requires a tuning
>> >>>>>>>>> parameter that, if to small, won't mitigate the problem, but if to
>> >>>>>>>>> big, greatly increases latency. (Possibly one could define a dynamic
>> >>>>>>>>> session-like window to solve this though...) It also might be harder
>> >>>>>>>>> for runners that *can* cheaply present stuff in timestamp order to
>> >>>>>>>>> optimize. (That and, in practice, our annotation-style process methods
>> >>>>>>>>> don't lend themselves to easy composition.) I think it could work in
>> >>>>>>>>> specific cases though.
>> >>>>>>>>>
>> >>>>>>>>> More inline below.
>> >>>>>>>>>
>> >>>>>>>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz> wrote:
>> >>>>>>>>>> Hi Robert,
>> >>>>>>>>>>
>> >>>>>>>>>> > Beam has an exactly-once model. If the data was consumed, state
>> >>>>>>>>>> mutated, and outputs written downstream (these three are committed
>> >>>>>>>>>> together atomically) it will not be replayed. That does not, of
>> >>>>>>>>>> course,
>> >>>>>>>>>> solve the non-determanism due to ordering (including the fact that two
>> >>>>>>>>>> operations reading the same PCollection may view different ordering).
>> >>>>>>>>>>
>> >>>>>>>>>> I think what you describe is a property of a runner, not of the model,
>> >>>>>>>>>> right? I think if I run my pipeline on Flink I will not get this
>> >>>>>>>>>> atomicity, because although Flink uses also exactly-once model if
>> >>>>>>>>>> might
>> >>>>>>>>>> write outputs multiple times.
>> >>>>>>>>> Actually, I think it is a larger (open) question whether exactly once
>> >>>>>>>>> is guaranteed by the model or whether runners are allowed to relax
>> >>>>>>>>> that. I would think, however, that sources correctly implemented
>> >>>>>>>>> should be idempotent when run atop an exactly once infrastructure such
>> >>>>>>>>> as Flink of Dataflow.
>> >>>>>>>>>
>> >>>>>>>>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
>> >>>>>>>>>> received
>> >>>>>>>>>> in a specific order? In the current model, it is not. Being able to
>> >>>>>>>>>> read, handle, and produced out-of-order data, including late data,
>> >>>>>>>>>> is a
>> >>>>>>>>>> pretty fundamental property of distributed systems.
>> >>>>>>>>>>
>> >>>>>>>>>> Yes, absolutely. The argument here is not that Stateful ParDo should
>> >>>>>>>>>> presume to receive elements in any order, but to _present_ it as
>> >>>>>>>>>> such to
>> >>>>>>>>>> the user @ProcessElement function.
>> >>>>>>>>> Sounds like we should make this clearer.
>> >>>>>>>>>
>> >>>>>>>>>> > 2) Given that some operations are easier (or possibly only
>> >>>>>>>>>> possible)
>> >>>>>>>>>> to write when operating on ordered data, and that different runners
>> >>>>>>>>>> may
>> >>>>>>>>>> have (significantly) cheaper ways to provide this ordering than can be
>> >>>>>>>>>> done by the user themselves, should we elevate this to a property of
>> >>>>>>>>>> (Stateful?)DoFns that the runner can provide? I think a compelling
>> >>>>>>>>>> argument can be made here that we should.
>> >>>>>>>>>>
>> >>>>>>>>>> +1
>> >>>>>>>>>>
>> >>>>>>>>>> Jan
>> >>>>>>>>>>
>> >>>>>>>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>> >>>>>>>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz> wrote:
>> >>>>>>>>>>>> > I don't see batch vs. streaming as part of the model. One
>> >>>>>>>>>>>> can have
>> >>>>>>>>>>>> microbatch, or even a runner that alternates between different
>> >>>>>>>>>>>> modes.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Although I understand motivation of this statement, this project
>> >>>>>>>>>>>> name is
>> >>>>>>>>>>>> "Apache Beam: An advanced unified programming model". What does the
>> >>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>> >>>>>>>>>>> What I mean is that streaming vs. batch is no longer part of the
>> >>>>>>>>>>> model
>> >>>>>>>>>>> (or ideally API), but pushed down to be a concern of the runner
>> >>>>>>>>>>> (executor) of the pipeline.
>> >>>>>>>>>>>
>> >>>>>>>>>>>
>> >>>>>>>>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz>
>> >>>>>>>>>>> wrote:
>> >>>>>>>>>>>> Hi Kenn,
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> OK, so if we introduce annotation, we can have stateful ParDo
>> >>>>>>>>>>>> with sorting, that would perfectly resolve my issues. I still
>> >>>>>>>>>>>> have some doubts, though. Let me explain. The current behavior of
>> >>>>>>>>>>>> stateful ParDo has the following properties:
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> a) might fail in batch, although runs fine in streaming (that
>> >>>>>>>>>>>> is due to the buffering, and unbounded lateness in batch, which
>> >>>>>>>>>>>> was discussed back and forth in this thread)
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> b) might be non deterministic (this is because the elements
>> >>>>>>>>>>>> arrive at somewhat random order, and even if you do the operation
>> >>>>>>>>>>>> "assign unique ID to elements" this might produce different
>> >>>>>>>>>>>> results when run multiple times)
>> >>>>>>>>>>> PCollections are *explicitly* unordered. Any operations that
>> >>>>>>>>>>> assume or
>> >>>>>>>>>>> depend on a specific ordering for correctness (or determinism) must
>> >>>>>>>>>>> provide that ordering themselves (i.e. tolerate "arbitrary shuffling
>> >>>>>>>>>>> of inputs"). As you point out, that may be very expensive if you have
>> >>>>>>>>>>> very hot keys with very large (unbounded) timestamp skew.
>> >>>>>>>>>>>
>> >>>>>>>>>>> StatefulDoFns are low-level operations that should be used with care;
>> >>>>>>>>>>> the simpler windowing model gives determinism in the face of
>> >>>>>>>>>>> unordered
>> >>>>>>>>>>> data (though late data and non-end-of-window triggering introduces
>> >>>>>>>>>>> some of the non-determanism back in).
>> >>>>>>>>>>>
>> >>>>>>>>>>>> What worries me most is the property b), because it seems to me
>> >>>>>>>>>>>> to have serious consequences - not only that if you run twice
>> >>>>>>>>>>>> batch pipeline you would get different results, but even on
>> >>>>>>>>>>>> streaming, when pipeline fails and gets restarted from
>> >>>>>>>>>>>> checkpoint, produced output might differ from the previous run
>> >>>>>>>>>>>> and data from the first run might have already been persisted
>> >>>>>>>>>>>> into sink. That would create somewhat messy outputs.
>> >>>>>>>>>>> Beam has an exactly-once model. If the data was consumed, state
>> >>>>>>>>>>> mutated, and outputs written downstream (these three are committed
>> >>>>>>>>>>> together atomically) it will not be replayed. That does not, of
>> >>>>>>>>>>> course, solve the non-determanism due to ordering (including the fact
>> >>>>>>>>>>> that two operations reading the same PCollection may view different
>> >>>>>>>>>>> ordering).
>> >>>>>>>>>>>
>> >>>>>>>>>>>> These two properties makes me think that the current
>> >>>>>>>>>>>> implementation is more of a _special case_ than the general one.
>> >>>>>>>>>>>> The general one would be that your state doesn't have the
>> >>>>>>>>>>>> properties to be able to tolerate buffering problems and/or
>> >>>>>>>>>>>> non-determinism. Which is the case where you need sorting in both
>> >>>>>>>>>>>> streaming and batch to be part of the model.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Let me point out one more analogy - that is merging vs.
>> >>>>>>>>>>>> non-merging windows. The general case (merging windows) implies
>> >>>>>>>>>>>> sorting by timestamp in both batch case (explicit) and streaming
>> >>>>>>>>>>>> (buffering). The special case (non-merging windows) doesn't rely
>> >>>>>>>>>>>> on any timestamp ordering, so the sorting and buffering can be
>> >>>>>>>>>>>> dropped. The underlying root cause of this is the same for both
>> >>>>>>>>>>>> stateful ParDo and windowing (essentially, assigning window
>> >>>>>>>>>>>> labels is a stateful operation when windowing function is merging).
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> The reason for the current behavior of stateful ParDo seems to be
>> >>>>>>>>>>>> performance, but is it right to abandon correctness in favor of
>> >>>>>>>>>>>> performance? Wouldn't it be more consistent to have the default
>> >>>>>>>>>>>> behavior prefer correctness and when you have the specific
>> >>>>>>>>>>>> conditions of state function having special properties, then you
>> >>>>>>>>>>>> can annotate your DoFn (with something like
>> >>>>>>>>>>>> @TimeOrderingAgnostic), which would yield a better performance in
>> >>>>>>>>>>>> that case?
>> >>>>>>>>>>> There are two separable questions here.
>> >>>>>>>>>>>
>> >>>>>>>>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are received
>> >>>>>>>>>>> in a specific order? In the current model, it is not. Being able to
>> >>>>>>>>>>> read, handle, and produced out-of-order data, including late data, is
>> >>>>>>>>>>> a pretty fundamental property of distributed systems.
>> >>>>>>>>>>>
>> >>>>>>>>>>> 2) Given that some operations are easier (or possibly only possible)
>> >>>>>>>>>>> to write when operating on ordered data, and that different runners
>> >>>>>>>>>>> may have (significantly) cheaper ways to provide this ordering than
>> >>>>>>>>>>> can be done by the user themselves, should we elevate this to a
>> >>>>>>>>>>> property of (Stateful?)DoFns that the runner can provide? I think a
>> >>>>>>>>>>> compelling argument can be made here that we should.
>> >>>>>>>>>>>
>> >>>>>>>>>>> - Robert
>> >>>>>>>>>>>
>> >>>>>>>>>>>
>> >>>>>>>>>>>
>> >>>>>>>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Thanks for the nice small example of a calculation that depends
>> >>>>>>>>>>>> on order. You are right that many state machines have this
>> >>>>>>>>>>>> property. I agree w/ you and Luke that it is convenient for batch
>> >>>>>>>>>>>> processing to sort by event timestamp before running a stateful
>> >>>>>>>>>>>> ParDo. In streaming you could also implement "sort by event
>> >>>>>>>>>>>> timestamp" by buffering until you know all earlier data will be
>> >>>>>>>>>>>> dropped - a slack buffer up to allowed lateness.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> I do not think that it is OK to sort in batch and not in
>> >>>>>>>>>>>> streaming. Many state machines diverge very rapidly when things
>> >>>>>>>>>>>> are out of order. So each runner if they see the
>> >>>>>>>>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
>> >>>>>>>>>>>> sorted data (by some mix of buffering and dropping), or to reject
>> >>>>>>>>>>>> the pipeline as unsupported.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> And also want to say that this is not the default case - many
>> >>>>>>>>>>>> uses of state & timers in ParDo yield different results at the
>> >>>>>>>>>>>> element level, but the results are equivalent at in the big
>> >>>>>>>>>>>> picture. Such as the example of "assign a unique sequence number
>> >>>>>>>>>>>> to each element" or "group into batches" it doesn't matter
>> >>>>>>>>>>>> exactly what the result is, only that it meets the spec. And
>> >>>>>>>>>>>> other cases like user funnels are monotonic enough that you also
>> >>>>>>>>>>>> don't actually need sorting.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Kenn
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz>
>> >>>>>>>>>>>> wrote:
>> >>>>>>>>>>>>> Yes, the problem will arise probably mostly when you have not
>> >>>>>>>>>>>>> well distributed keys (or too few keys). I'm really not sure if
>> >>>>>>>>>>>>> a pure GBK with a trigger can solve this - it might help to have
>> >>>>>>>>>>>>> data driven trigger. There would still be some doubts, though.
>> >>>>>>>>>>>>> The main question is still here - people say, that sorting by
>> >>>>>>>>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
>> >>>>>>>>>>>>> I don't really see why - the sorting is very probably already
>> >>>>>>>>>>>>> there. And if not (hash grouping instead of sorted grouping),
>> >>>>>>>>>>>>> then the sorting would affect only user defined StatefulParDos.
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> This would suggest that the best way out of this would be really
>> >>>>>>>>>>>>> to add annotation, so that the author of the pipeline can decide.
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> If that would be acceptable I think I can try to prepare some
>> >>>>>>>>>>>>> basic functionality, but I'm not sure, if I would be able to
>> >>>>>>>>>>>>> cover all runners / sdks.
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> It is read all per key and window and not just read all (this
>> >>>>>>>>>>>>> still won't scale with hot keys in the global window). The GBK
>> >>>>>>>>>>>>> preceding the StatefulParDo will guarantee that you are
>> >>>>>>>>>>>>> processing all the values for a specific key and window at any
>> >>>>>>>>>>>>> given time. Is there a specific window/trigger that is missing
>> >>>>>>>>>>>>> that you feel would remove the need for you to use StatefulParDo?
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
>> >>>>>>>>>>>>> wrote:
>> >>>>>>>>>>>>>> Hi Lukasz,
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Today, if you must have a strict order, you must guarantee
>> >>>>>>>>>>>>>>> that your StatefulParDo implements the necessary "buffering &
>> >>>>>>>>>>>>>>> sorting" into state.
>> >>>>>>>>>>>>>> Yes, no problem with that. But this whole discussion started,
>> >>>>>>>>>>>>>> because *this doesn't work on batch*. You simply cannot first
>> >>>>>>>>>>>>>> read everything from distributed storage and then buffer it all
>> >>>>>>>>>>>>>> into memory, just to read it again, but sorted. That will not
>> >>>>>>>>>>>>>> work. And even if it would, it would be a terrible waste of
>> >>>>>>>>>>>>>> resources.
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> Jan
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
>> >>>>>>>>>>>>>> wrote:
>> >>>>>>>>>>>>>>> This discussion brings many really interesting questions for
>> >>>>>>>>>>>>>>> me. :-)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > I don't see batch vs. streaming as part of the model. One
>> >>>>>>>>>>>>>>> can have
>> >>>>>>>>>>>>>>> microbatch, or even a runner that alternates between different
>> >>>>>>>>>>>>>>> modes.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Although I understand motivation of this statement, this
>> >>>>>>>>>>>>>>> project name is
>> >>>>>>>>>>>>>>> "Apache Beam: An advanced unified programming model". What
>> >>>>>>>>>>>>>>> does the
>> >>>>>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
>> >>>>>>>>>>>>>>> are
>> >>>>>>>>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
>> >>>>>>>>>>>>>>> All these
>> >>>>>>>>>>>>>>> define several runtime parameters, which in turn define how
>> >>>>>>>>>>>>>>> well/badly
>> >>>>>>>>>>>>>>> will the pipeline perform and how many resources might be
>> >>>>>>>>>>>>>>> needed. From
>> >>>>>>>>>>>>>>> my point of view, pure streaming should be the most resource
>> >>>>>>>>>>>>>>> demanding
>> >>>>>>>>>>>>>>> (if not, why bother with batch? why not run everything in
>> >>>>>>>>>>>>>>> streaming
>> >>>>>>>>>>>>>>> only? what will there remain to "unify"?).
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Fortunately, for batch, only the state for a single key
>> >>>>>>>>>>>>>>> needs to be
>> >>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys across
>> >>>>>>>>>>>>>>> the range
>> >>>>>>>>>>>>>>> of skew. Of course if you have few or hot keys, one can still
>> >>>>>>>>>>>>>>> have
>> >>>>>>>>>>>>>>> issues (and this is not specific to StatefulDoFns).
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Yes, but here is still the presumption that my stateful DoFn can
>> >>>>>>>>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
>> >>>>>>>>>>>>>>> case in
>> >>>>>>>>>>>>>>> more detail.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
>> >>>>>>>>>>>>>>> some key for
>> >>>>>>>>>>>>>>> each element, which is irrelevant for the demonstration). Your
>> >>>>>>>>>>>>>>> task is
>> >>>>>>>>>>>>>>> to calculate in running global window the actual number of
>> >>>>>>>>>>>>>>> changes
>> >>>>>>>>>>>>>>> between state 0 and state 1 and vice versa. When the state
>> >>>>>>>>>>>>>>> doesn't
>> >>>>>>>>>>>>>>> change, you don't calculate anything. If input (for given key)
>> >>>>>>>>>>>>>>> would be
>> >>>>>>>>>>>>>>> (tN denotes timestamp N):
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t1: 1
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t2: 0
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t3: 0
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t4: 1
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t5: 1
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t6: 0
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> then the output should yield (supposing that default state is
>> >>>>>>>>>>>>>>> zero):
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t1: (one: 1, zero: 0)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t2: (one: 1, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t3: (one: 1, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t4: (one: 2, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t5: (one: 2, zero: 1)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> t6: (one: 2, zero: 2)
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> How would you implement this in current Beam semantics?
>> >>>>>>>>>>>>>> I think your saying here that I know that my input is ordered
>> >>>>>>>>>>>>>> in a specific way and since I assume the order when writing my
>> >>>>>>>>>>>>>> pipeline I can perform this optimization. But there is nothing
>> >>>>>>>>>>>>>> preventing a runner from noticing that your processing in the
>> >>>>>>>>>>>>>> global window with a specific type of trigger and re-ordering
>> >>>>>>>>>>>>>> your inputs/processing to get better performance (since you
>> >>>>>>>>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>> >>>>>>>>>>>>>> streaming for the GlobalWindow).
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>> Today, if you must have a strict order, you must guarantee that
>> >>>>>>>>>>>>>> your StatefulParDo implements the necessary "buffering &
>> >>>>>>>>>>>>>> sorting" into state. I can see why you would want an annotation
>> >>>>>>>>>>>>>> that says I must have timestamp ordered elements, since it
>> >>>>>>>>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
>> >>>>>>>>>>>>>> is a low-level function, it really is the "here you go and do
>> >>>>>>>>>>>>>> whatever you need to but here be dragons" function while
>> >>>>>>>>>>>>>> windowing and triggering is meant to keep many people from
>> >>>>>>>>>>>>>> writing StatefulParDo in the first place.
>> >>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
>> >>>>>>>>>>>>>>> are likely to
>> >>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>> >>>>>>>>>>>>>>> falls
>> >>>>>>>>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> But the worst case is defined by input of size (available
>> >>>>>>>>>>>>>>> resources +
>> >>>>>>>>>>>>>>> single byte) -> pipeline fail. Although it could have
>> >>>>>>>>>>>>>>> finished, given
>> >>>>>>>>>>>>>>> the right conditions.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > This might be reasonable, implemented by default by
>> >>>>>>>>>>>>>>> buffering
>> >>>>>>>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>> >>>>>>>>>>>>>>> advances,
>> >>>>>>>>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
>> >>>>>>>>>>>>>>> reason
>> >>>>>>>>>>>>>>> about) code.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Sure, the pipeline will be less efficient, because it would
>> >>>>>>>>>>>>>>> have to
>> >>>>>>>>>>>>>>> buffer and sort the inputs. But at least it will produce
>> >>>>>>>>>>>>>>> correct results
>> >>>>>>>>>>>>>>> in cases where updates to state are order-sensitive.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
>> >>>>>>>>>>>>>>> (key, values):
>> >>>>>>>>>>>>>>> [(key, value) for value in values])?
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
>> >>>>>>>>>>>>>>> The trigger
>> >>>>>>>>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
>> >>>>>>>>>>>>>>> element
>> >>>>>>>>>>>>>>> with lowest timestamp in the buffer. Although this could be
>> >>>>>>>>>>>>>>> somehow
>> >>>>>>>>>>>>>>> emulated by fixed trigger each X millis.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> > Or is the underlying desire just to be able to hint to
>> >>>>>>>>>>>>>>> the runner
>> >>>>>>>>>>>>>>> that the code may perform better (e.g. require less resources)
>> >>>>>>>>>>>>>>> as skew
>> >>>>>>>>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> No, the sorting would have to be done in streaming case as
>> >>>>>>>>>>>>>>> well. That is
>> >>>>>>>>>>>>>>> an imperative of the unified model. I think it is possible to
>> >>>>>>>>>>>>>>> sort by
>> >>>>>>>>>>>>>>> timestamp only in batch case (and do it for *all* batch
>> >>>>>>>>>>>>>>> stateful pardos
>> >>>>>>>>>>>>>>> without annotation), or introduce annotation, but then make
>> >>>>>>>>>>>>>>> the same
>> >>>>>>>>>>>>>>> guarantees for streaming case as well.
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> Jan
>> >>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>> >>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>> >>>>>>>>>>>>>>>>> Hi Robert,
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
>> >>>>>>>>>>>>>>>>> guarantees
>> >>>>>>>>>>>>>>>>> of ordering are given in either mode, there is *implicit*
>> >>>>>>>>>>>>>>>>> ordering in
>> >>>>>>>>>>>>>>>>> streaming case that is due to nature of the processing - the
>> >>>>>>>>>>>>>>>>> difference
>> >>>>>>>>>>>>>>>>> between watermark and timestamp of elements flowing through
>> >>>>>>>>>>>>>>>>> the pipeline
>> >>>>>>>>>>>>>>>>> are generally low (too high difference leads to the
>> >>>>>>>>>>>>>>>>> overbuffering
>> >>>>>>>>>>>>>>>>> problem), but there is no such bound on batch.
>> >>>>>>>>>>>>>>>> Fortunately, for batch, only the state for a single key needs
>> >>>>>>>>>>>>>>>> to be
>> >>>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys
>> >>>>>>>>>>>>>>>> across the
>> >>>>>>>>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
>> >>>>>>>>>>>>>>>> still
>> >>>>>>>>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> As a result, I see a few possible solutions:
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> - the best and most natural seems to be extension of
>> >>>>>>>>>>>>>>>>> the model, so
>> >>>>>>>>>>>>>>>>> that it defines batch as not only "streaming pipeline
>> >>>>>>>>>>>>>>>>> executed in batch
>> >>>>>>>>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>> >>>>>>>>>>>>>>>>> characteristics as
>> >>>>>>>>>>>>>>>>> in streaming case, executed in batch fashion", I really
>> >>>>>>>>>>>>>>>>> don't think that
>> >>>>>>>>>>>>>>>>> there are any conflicts with the current model, or that this
>> >>>>>>>>>>>>>>>>> could
>> >>>>>>>>>>>>>>>>> affect performance, because the required sorting (as pointed by
>> >>>>>>>>>>>>>>>>> Aljoscha) is very probably already done during translation
>> >>>>>>>>>>>>>>>>> of stateful
>> >>>>>>>>>>>>>>>>> pardos. Also note that this definition only affects user
>> >>>>>>>>>>>>>>>>> defined
>> >>>>>>>>>>>>>>>>> stateful pardos
>> >>>>>>>>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
>> >>>>>>>>>>>>>>>> have
>> >>>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
>> >>>>>>>>>>>>>>>> different modes.
>> >>>>>>>>>>>>>>>> The model describes what the valid outputs are given a
>> >>>>>>>>>>>>>>>> (sometimes
>> >>>>>>>>>>>>>>>> partial) set of inputs. It becomes really hard to define
>> >>>>>>>>>>>>>>>> things like
>> >>>>>>>>>>>>>>>> "as good runtime characteristics." Once you allow any
>> >>>>>>>>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
>> >>>>>>>>>>>>>>>> (and
>> >>>>>>>>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>> >>>>>>>>>>>>>>>> out-of-orderedness.
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
>> >>>>>>>>>>>>>>>> likely to
>> >>>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>> >>>>>>>>>>>>>>>> falls
>> >>>>>>>>>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> - another option would be to introduce annotation for
>> >>>>>>>>>>>>>>>>> DoFns (e.g.
>> >>>>>>>>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
>> >>>>>>>>>>>>>>>>> the sorting
>> >>>>>>>>>>>>>>>>> in batch case - but - this extension would have to ensure
>> >>>>>>>>>>>>>>>>> the sorting in
>> >>>>>>>>>>>>>>>>> streaming mode also - it would require definition of allowed
>> >>>>>>>>>>>>>>>>> lateness,
>> >>>>>>>>>>>>>>>>> and triggger (essentially similar to window)
>> >>>>>>>>>>>>>>>> This might be reasonable, implemented by default by buffering
>> >>>>>>>>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>> >>>>>>>>>>>>>>>> advances, but would likely lead to inefficient (though
>> >>>>>>>>>>>>>>>> *maybe* easier
>> >>>>>>>>>>>>>>>> to reason about) code. Not sure about the semantics of
>> >>>>>>>>>>>>>>>> triggering
>> >>>>>>>>>>>>>>>> here, especially data-driven triggers. Would it be roughly
>> >>>>>>>>>>>>>>>> equivalent
>> >>>>>>>>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
>> >>>>>>>>>>>>>>>> value in
>> >>>>>>>>>>>>>>>> values])?
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>> Or is the underlying desire just to be able to hint to the
>> >>>>>>>>>>>>>>>> runner that
>> >>>>>>>>>>>>>>>> the code may perform better (e.g. require less resources) as
>> >>>>>>>>>>>>>>>> skew is
>> >>>>>>>>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>> >>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> - last option would be to introduce these "higher order
>> >>>>>>>>>>>>>>>>> guarantees" in
>> >>>>>>>>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
>> >>>>>>>>>>>>>>>>> worst
>> >>>>>>>>>>>>>>>>> option to me
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> I see the first two options quite equally good, although the
>> >>>>>>>>>>>>>>>>> letter one
>> >>>>>>>>>>>>>>>>> is probably more time consuming to implement. But it would
>> >>>>>>>>>>>>>>>>> bring
>> >>>>>>>>>>>>>>>>> additional feature to streaming case as well.
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> Thanks for any thoughts.
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> Jan
>> >>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>> >>>>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>> >>>>>>>>>>>>>>>>>>> Hi Reuven,
>> >>>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
>> >>>>>>>>>>>>>>>>>>>> runners.
>> >>>>>>>>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
>> >>>>>>>>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
>> >>>>>>>>>>>>>>>>>>> of elements. That basically (practically) can work only
>> >>>>>>>>>>>>>>>>>>> for cases, where the order of input elements doesn't
>> >>>>>>>>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
>> >>>>>>>>>>>>>>>>>>> time you have a state machine involved, then the ordering
>> >>>>>>>>>>>>>>>>>>> of elements would matter.
>> >>>>>>>>>>>>>>>>>> No guarantees on order are provided in *either* streaming
>> >>>>>>>>>>>>>>>>>> or batch
>> >>>>>>>>>>>>>>>>>> mode by the model. However, it is the case that in order to
>> >>>>>>>>>>>>>>>>>> make
>> >>>>>>>>>>>>>>>>>> forward progress most streaming runners attempt to limit
>> >>>>>>>>>>>>>>>>>> the amount of
>> >>>>>>>>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
>> >>>>>>>>>>>>>>>>>> processing
>> >>>>>>>>>>>>>>>>>> time) to make forward progress, which in turn could help
>> >>>>>>>>>>>>>>>>>> cap the
>> >>>>>>>>>>>>>>>>>> amount of state that must be held concurrently, whereas a
>> >>>>>>>>>>>>>>>>>> batch runner
>> >>>>>>>>>>>>>>>>>> may not allow any state to be safely discarded until the whole
>> >>>>>>>>>>>>>>>>>> timeline from infinite past to infinite future has been
>> >>>>>>>>>>>>>>>>>> observed.
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
>> >>>>>>>>>>>>>>>>>> batch" in batch mode.
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>> >>>>>>>>>>>>>>>>>> <mx...@apache.org> wrote:
>> >>>>>>>>>>>>>>>>>>
>> >>>>>>>>>>>>>>>>>>>> batch semantics and streaming semantics differs only
>> >>>>>>>>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
>> >>>>>>>>>>>>>>>>>>>> batch and cannot on stream
>> >>>>>>>>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
>> >>>>>>>>>>>>>>>>>>> trigger. You
>> >>>>>>>>>>>>>>>>>>> could define additional triggers that do early firings.
>> >>>>>>>>>>>>>>>>>>> And you could
>> >>>>>>>>>>>>>>>>>>> even trigger the global window by advancing the watermark
>> >>>>>>>>>>>>>>>>>>> to +inf.
>> >>>>>>>>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
>> >>>>>>>>>>>>>>>>>> default
>> >>>>>>>>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
>> >>>>>>>>>>>>>>>>>> is more
>> >>>>>>>>>>>>>>>>>> likely to be user error than an actual desire to have no
>> >>>>>>>>>>>>>>>>>> output until
>> >>>>>>>>>>>>>>>>>> drain. But it's semantically valid in the model.
>>
Re: Definition of Unified model
Posted by Reuven Lax <re...@google.com>.
Sequence metadata does have the disadvantage that users can no longer use
the types coming from the source. You must create a new type that contains
a sequence number (unless Beam provides this). It also gets awkward with
Flatten - the sequence number is no longer enough, you must also encode
which side of the flatten each element came from.
On Tue, May 28, 2019 at 3:18 AM Jan Lukavský <je...@seznam.cz> wrote:
> As I understood it, Kenn was supporting the idea that sequence metadata
> is preferable over FIFO. I was trying to point out, that it even should
> provide the same functionally as FIFO, plus one important more -
> reproducibility and ability to being persisted and reused the same way
> in batch and streaming.
>
> There is no doubt, that sequence metadata can be stored in every
> storage. But, regarding some implicit ordering that sources might have -
> yes, of course, data written into HDFS or Cloud Storage has ordering,
> but only partial - inside some bulk (e.g. file) and the ordering is not
> defined correctly on boundaries of these bulks (between files). That is
> why I'd say, that ordering of sources is relevant only for
> (partitioned!) streaming sources and generally always reduces to
> sequence metadata (e.g. offsets).
>
> Jan
>
> On 5/28/19 11:43 AM, Robert Bradshaw wrote:
> > Huge +1 to all Kenn said.
> >
> > Jan, batch sources can have orderings too, just like Kafka. I think
> > it's reasonable (for both batch and streaming) that if a source has an
> > ordering that is an important part of the data, it should preserve
> > this ordering into the data itself (e.g. as sequence numbers, offsets,
> > etc.)
> >
> > On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org>
> wrote:
> >> I strongly prefer explicit sequence metadata over FIFO requirements,
> because:
> >>
> >> - FIFO is complex to specify: for example Dataflow has "per stage
> key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage"
> is not a portable concept, nor even guaranteed to remain a Dataflow concept)
> >> - complex specifications are by definition poor usability (if
> necessary, then it is what it is)
> >> - overly restricts the runner, reduces parallelism, for example any
> non-stateful ParDo has per-element parallelism, not per "key"
> >> - another perspective on that: FIFO makes everyone pay rather than
> just the transform that requires exactly sequencing
> >> - previous implementation details like reshuffles become part of the
> model
> >> - I'm not even convinced the use cases involved are addressed by some
> careful FIFO restrictions; many sinks re-key and they would all have to
> become aware of how keying of a sequence of "stages" affects the end-to-end
> FIFO
> >>
> >> A noop becoming a non-noop is essentially the mathematical definition
> of moving from higher-level to lower-level abstraction.
> >>
> >> So this strikes at the core question of what level of abstraction Beam
> aims to represent. Lower-level means there are fewer possible
> implementations and it is more tied to the underlying architecture, and
> anything not near-exact match pays a huge penalty. Higher-level means there
> are more implementations possible with different tradeoffs, though they may
> all pay a minor penalty.
> >>
> >> I could be convinced to change my mind, but it needs some extensive
> design, examples, etc. I think it is probably about the most consequential
> design decision in the whole Beam model, around the same level as the
> decision to use ParDo and GBK as the primitives IMO.
> >>
> >> Kenn
> >>
> >> On Thu, May 23, 2019 at 10:17 AM Reuven Lax <re...@google.com> wrote:
> >>> Not really. I'm suggesting that some variant of FIFO ordering is
> necessary, which requires either runners natively support FIFO ordering or
> transforms adding some extra sequence number to each record to sort by.
> >>>
> >>> I still think your proposal is very useful by the way. I'm merely
> pointing out that to solve the state-machine problem we probably need
> something more.
> >>>
> >>> Reuven
> >>>
> >>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je...@seznam.cz> wrote:
> >>>> Hi,
> >>>> yes. It seems that ordering by user supplied UDF makes sense and I
> will update the design proposal accordingly.
> >>>> Would that solve the issues you mention?
> >>>> Jan
> >>>> ---------- Původní e-mail ----------
> >>>> Od: Reuven Lax <re...@google.com>
> >>>> Komu: dev <de...@beam.apache.org>
> >>>> Datum: 23. 5. 2019 18:44:38
> >>>> Předmět: Re: Definition of Unified model
> >>>>
> >>>> I'm simply saying that timestamp ordering is insufficient for state
> machines. I wasn't proposing Kafka as a solution - that was simply an
> example of how people solve this problem in other scenarios.
> >>>>
> >>>> BTW another example of ordering: Imagine today that you have a
> triggered Sum aggregation writing out to a key-value sink. In theory we
> provide no ordering, so the sink might write the triggered sums in the
> wrong order, ending up with an incorrect value in the sink. In this case
> you probably want values ordered by trigger pane index.
> >>>>
> >>>> Reuven
> >>>>
> >>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz> wrote:
> >>>>
> >>>> Hi Reuven,
> >>>> I share the view point of Robert. I think the isuue you refer to is
> not in reality related to timestamps, but to the fact, that ordering of
> events in time is observer dependent (either caused by relativity, or time
> skew, essentially this has the same consequences). And the resolution in
> fact isn't Kafka, but generally an authoritative observer, that tells you
> "I saw the events in this order". And you either have one (and have the
> outcome of his observation persisted in the data - e.g. as offset in Kafka
> partition), then you should be able to use it (maybe that suggests afterall
> that sorting by some user supplied UDF might make sense), or do not have
> it, and then any interpretation of the data seems to be equally valid.
> Although determinism is fine, of course.
> >>>> Jan
> >>>> ---------- Původní e-mail ----------
> >>>> Od: Reuven Lax <re...@google.com>
> >>>> Komu: dev <de...@beam.apache.org>
> >>>> Datum: 23. 5. 2019 17:39:12
> >>>> Předmět: Re: Definition of Unified model
> >>>>
> >>>> So an example would be elements of type "startUserSession" and
> "endUserSession" (website sessions, not Beam sessions). Logically you may
> need to process them in the correct order if you have any sort of
> state-machine logic. However timestamp ordering is never guaranteed to
> match the logical ordering. Not only might you have several elements with
> the same timestamp, but in reality time skew across backend servers can
> cause the events to have timestamps in reverse order of the actual
> causality order.
> >>>>
> >>>> People do solve this problem today though. Publish the events to
> Kafka, making sure that events for the same user end up in the same Kafka
> partition. This ensures that the events appear in the Kafka partitions in
> causality order, even if the timestamp order doesn't match. The your Kafka
> subscriber simply process the elements in each partition in order.
> >>>>
> >>>> I think the ability to impose FIFO causality ordering is what's
> needed for any state-machine work. Timestamp ordering has advantages
> (though often I think the advantage is in state), but does not solve this
> problem.
> >>>>
> >>>> Reuven
> >>>>
> >>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <ro...@google.com>
> wrote:
> >>>>
> >>>> Good point.
> >>>>
> >>>> The "implementation-specific" way I would do this is
> >>>> window-by-instant, followed by a DoFn that gets all the elements with
> >>>> the same timestamp and sorts/acts accordingly, but this counts on the
> >>>> runner producing windows in timestamp order (likely?) and also the
> >>>> subsequent DoFn getting them in this order (also likely, due to
> >>>> fusion).
> >>>>
> >>>> One could make the argument that, though it does not provide
> >>>> deterministic behavior, getting elements of the same timestamp in
> >>>> different orders should produce equally valid interpretations of the
> >>>> data. (After all, due to relatively, timestamps are not technically
> >>>> well ordered across space.) I can see how data-dependent tiebreakers
> >>>> could be useful, or promises of preservation of order between
> >>>> operations.
> >>>>
> >>>> - Robert
> >>>>
> >>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com> wrote:
> >>>>> So Jan's example of state machines is quite a valid use case for
> ordering. However in my experience, timestamp ordering is insufficient for
> state machines. Elements that cause state transitions might come in with
> the exact same timestamp, yet still have a necessary ordering. Especially
> given Beam's decision to have milliseconds timestamps this is possible, but
> even at microsecond or nanosecond precision this can happen at scale. To
> handle state machines you usually need some sort of FIFO ordering along
> with an ordered sources, such as Kafka, not timestamp ordering.
> >>>>>
> >>>>> Reuven
> >>>>>
> >>>>> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>> Hi all,
> >>>>>>
> >>>>>> thanks everyone for this discussion. I think I have gathered enough
> >>>>>> feedback to be able to put down a proposition for changes, which I
> will
> >>>>>> do and send to this list for further discussion. There are still
> doubts
> >>>>>> remaining the non-determinism and it's relation to outputs
> stability vs.
> >>>>>> latency. But I will try to clarify all this in the design document.
> >>>>>>
> >>>>>> Thanks,
> >>>>>>
> >>>>>> Jan
> >>>>>>
> >>>>>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
> >>>>>>>> current understanding.
> >>>>>>> In essence your description of how exactly-once works in Flink is
> >>>>>>> correct. The general assumption in Flink is that pipelines must be
> >>>>>>> deterministic and thus produce idempotent writes in the case of
> >>>>>>> failures. However, that doesn't mean Beam sinks can't guarantee a
> bit
> >>>>>>> more with what Flink has to offer.
> >>>>>>>
> >>>>>>> Luke already mentioned the design discussions for
> @RequiresStableInput
> >>>>>>> which ensures idempotent writes for non-deterministic pipelines.
> This
> >>>>>>> is not part of the model but an optional Beam feature.
> >>>>>>>
> >>>>>>> We recently implemented support for @RequiresStableInput in the
> Flink
> >>>>>>> Runner. Reuven mentioned the Flink checkpoint confirmation, which
> >>>>>>> allows us to buffer (and checkpoint) processed data and only emit
> it
> >>>>>>> once a Flink checkpoint has completed.
> >>>>>>>
> >>>>>>> Cheers,
> >>>>>>> Max
> >>>>>>>
> >>>>>>> On 21.05.19 16:49, Jan Lukavský wrote:
> >>>>>>>> Hi,
> >>>>>>>>
> >>>>>>>> > Actually, I think it is a larger (open) question whether
> exactly
> >>>>>>>> once is guaranteed by the model or whether runners are allowed to
> >>>>>>>> relax that. I would think, however, that sources correctly
> >>>>>>>> implemented should be idempotent when run atop an exactly once
> >>>>>>>> infrastructure such as Flink of Dataflow.
> >>>>>>>>
> >>>>>>>> I would assume, that the model basically inherits guarantees of
> >>>>>>>> underlying infrastructure. Because Flink does not work as you
> >>>>>>>> described (atomic commit of inputs, state and outputs), but
> rather a
> >>>>>>>> checkpoint mark is flowing through the DAG much like watermark
> and on
> >>>>>>>> failures operators are restored and data reprocessed, it (IMHO)
> >>>>>>>> implies, that you have exactly once everywhere in the DAG *but*
> >>>>>>>> sinks. That is because sinks cannot be restored to previous state,
> >>>>>>>> instead sinks are supposed to be idempotent in order for the
> exactly
> >>>>>>>> once to really work (or at least be able to commit outputs on
> >>>>>>>> checkpoint in sink). That implies that if you don't have sink
> that is
> >>>>>>>> able to commit outputs atomically on checkpoint, the pipeline
> >>>>>>>> execution should be deterministic upon retries, otherwise shadow
> >>>>>>>> writes from failed paths of the pipeline might appear.
> >>>>>>>>
> >>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
> >>>>>>>> current understanding.
> >>>>>>>>
> >>>>>>>> > Sounds like we should make this clearer.
> >>>>>>>>
> >>>>>>>> I meant that you are right that we must not in any thoughts we are
> >>>>>>>> having forget that streams are by definition out-of-order. That is
> >>>>>>>> property that we cannot change. But - that doesn't limit us from
> >>>>>>>> creating operator that presents the data to UDF as if the stream
> was
> >>>>>>>> ideally sorted. It can do that by introducing latency, of course.
> >>>>>>>>
> >>>>>>>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
> >>>>>>>>> Reza: One could provide something like this as a utility class,
> but
> >>>>>>>>> one downside is that it is not scale invariant. It requires a
> tuning
> >>>>>>>>> parameter that, if to small, won't mitigate the problem, but if
> to
> >>>>>>>>> big, greatly increases latency. (Possibly one could define a
> dynamic
> >>>>>>>>> session-like window to solve this though...) It also might be
> harder
> >>>>>>>>> for runners that *can* cheaply present stuff in timestamp order
> to
> >>>>>>>>> optimize. (That and, in practice, our annotation-style process
> methods
> >>>>>>>>> don't lend themselves to easy composition.) I think it could
> work in
> >>>>>>>>> specific cases though.
> >>>>>>>>>
> >>>>>>>>> More inline below.
> >>>>>>>>>
> >>>>>>>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>>>>>> Hi Robert,
> >>>>>>>>>>
> >>>>>>>>>> > Beam has an exactly-once model. If the data was consumed,
> state
> >>>>>>>>>> mutated, and outputs written downstream (these three are
> committed
> >>>>>>>>>> together atomically) it will not be replayed. That does not, of
> >>>>>>>>>> course,
> >>>>>>>>>> solve the non-determanism due to ordering (including the fact
> that two
> >>>>>>>>>> operations reading the same PCollection may view different
> ordering).
> >>>>>>>>>>
> >>>>>>>>>> I think what you describe is a property of a runner, not of the
> model,
> >>>>>>>>>> right? I think if I run my pipeline on Flink I will not get this
> >>>>>>>>>> atomicity, because although Flink uses also exactly-once model
> if
> >>>>>>>>>> might
> >>>>>>>>>> write outputs multiple times.
> >>>>>>>>> Actually, I think it is a larger (open) question whether exactly
> once
> >>>>>>>>> is guaranteed by the model or whether runners are allowed to
> relax
> >>>>>>>>> that. I would think, however, that sources correctly implemented
> >>>>>>>>> should be idempotent when run atop an exactly once
> infrastructure such
> >>>>>>>>> as Flink of Dataflow.
> >>>>>>>>>
> >>>>>>>>>> > 1) Is it correct for a (Stateful)DoFn to assume elements
> are
> >>>>>>>>>> received
> >>>>>>>>>> in a specific order? In the current model, it is not. Being
> able to
> >>>>>>>>>> read, handle, and produced out-of-order data, including late
> data,
> >>>>>>>>>> is a
> >>>>>>>>>> pretty fundamental property of distributed systems.
> >>>>>>>>>>
> >>>>>>>>>> Yes, absolutely. The argument here is not that Stateful ParDo
> should
> >>>>>>>>>> presume to receive elements in any order, but to _present_ it as
> >>>>>>>>>> such to
> >>>>>>>>>> the user @ProcessElement function.
> >>>>>>>>> Sounds like we should make this clearer.
> >>>>>>>>>
> >>>>>>>>>> > 2) Given that some operations are easier (or possibly only
> >>>>>>>>>> possible)
> >>>>>>>>>> to write when operating on ordered data, and that different
> runners
> >>>>>>>>>> may
> >>>>>>>>>> have (significantly) cheaper ways to provide this ordering than
> can be
> >>>>>>>>>> done by the user themselves, should we elevate this to a
> property of
> >>>>>>>>>> (Stateful?)DoFns that the runner can provide? I think a
> compelling
> >>>>>>>>>> argument can be made here that we should.
> >>>>>>>>>>
> >>>>>>>>>> +1
> >>>>>>>>>>
> >>>>>>>>>> Jan
> >>>>>>>>>>
> >>>>>>>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
> >>>>>>>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>>>>>>>> > I don't see batch vs. streaming as part of the model.
> One
> >>>>>>>>>>>> can have
> >>>>>>>>>>>> microbatch, or even a runner that alternates between different
> >>>>>>>>>>>> modes.
> >>>>>>>>>>>>
> >>>>>>>>>>>> Although I understand motivation of this statement, this
> project
> >>>>>>>>>>>> name is
> >>>>>>>>>>>> "Apache Beam: An advanced unified programming model". What
> does the
> >>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the
> model?
> >>>>>>>>>>> What I mean is that streaming vs. batch is no longer part of
> the
> >>>>>>>>>>> model
> >>>>>>>>>>> (or ideally API), but pushed down to be a concern of the runner
> >>>>>>>>>>> (executor) of the pipeline.
> >>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je.ik@seznam.cz
> >
> >>>>>>>>>>> wrote:
> >>>>>>>>>>>> Hi Kenn,
> >>>>>>>>>>>>
> >>>>>>>>>>>> OK, so if we introduce annotation, we can have stateful ParDo
> >>>>>>>>>>>> with sorting, that would perfectly resolve my issues. I still
> >>>>>>>>>>>> have some doubts, though. Let me explain. The current
> behavior of
> >>>>>>>>>>>> stateful ParDo has the following properties:
> >>>>>>>>>>>>
> >>>>>>>>>>>> a) might fail in batch, although runs fine in streaming
> (that
> >>>>>>>>>>>> is due to the buffering, and unbounded lateness in batch,
> which
> >>>>>>>>>>>> was discussed back and forth in this thread)
> >>>>>>>>>>>>
> >>>>>>>>>>>> b) might be non deterministic (this is because the
> elements
> >>>>>>>>>>>> arrive at somewhat random order, and even if you do the
> operation
> >>>>>>>>>>>> "assign unique ID to elements" this might produce different
> >>>>>>>>>>>> results when run multiple times)
> >>>>>>>>>>> PCollections are *explicitly* unordered. Any operations that
> >>>>>>>>>>> assume or
> >>>>>>>>>>> depend on a specific ordering for correctness (or determinism)
> must
> >>>>>>>>>>> provide that ordering themselves (i.e. tolerate "arbitrary
> shuffling
> >>>>>>>>>>> of inputs"). As you point out, that may be very expensive if
> you have
> >>>>>>>>>>> very hot keys with very large (unbounded) timestamp skew.
> >>>>>>>>>>>
> >>>>>>>>>>> StatefulDoFns are low-level operations that should be used
> with care;
> >>>>>>>>>>> the simpler windowing model gives determinism in the face of
> >>>>>>>>>>> unordered
> >>>>>>>>>>> data (though late data and non-end-of-window triggering
> introduces
> >>>>>>>>>>> some of the non-determanism back in).
> >>>>>>>>>>>
> >>>>>>>>>>>> What worries me most is the property b), because it seems to
> me
> >>>>>>>>>>>> to have serious consequences - not only that if you run twice
> >>>>>>>>>>>> batch pipeline you would get different results, but even on
> >>>>>>>>>>>> streaming, when pipeline fails and gets restarted from
> >>>>>>>>>>>> checkpoint, produced output might differ from the previous run
> >>>>>>>>>>>> and data from the first run might have already been persisted
> >>>>>>>>>>>> into sink. That would create somewhat messy outputs.
> >>>>>>>>>>> Beam has an exactly-once model. If the data was consumed, state
> >>>>>>>>>>> mutated, and outputs written downstream (these three are
> committed
> >>>>>>>>>>> together atomically) it will not be replayed. That does not, of
> >>>>>>>>>>> course, solve the non-determanism due to ordering (including
> the fact
> >>>>>>>>>>> that two operations reading the same PCollection may view
> different
> >>>>>>>>>>> ordering).
> >>>>>>>>>>>
> >>>>>>>>>>>> These two properties makes me think that the current
> >>>>>>>>>>>> implementation is more of a _special case_ than the general
> one.
> >>>>>>>>>>>> The general one would be that your state doesn't have the
> >>>>>>>>>>>> properties to be able to tolerate buffering problems and/or
> >>>>>>>>>>>> non-determinism. Which is the case where you need sorting in
> both
> >>>>>>>>>>>> streaming and batch to be part of the model.
> >>>>>>>>>>>>
> >>>>>>>>>>>> Let me point out one more analogy - that is merging vs.
> >>>>>>>>>>>> non-merging windows. The general case (merging windows)
> implies
> >>>>>>>>>>>> sorting by timestamp in both batch case (explicit) and
> streaming
> >>>>>>>>>>>> (buffering). The special case (non-merging windows) doesn't
> rely
> >>>>>>>>>>>> on any timestamp ordering, so the sorting and buffering can be
> >>>>>>>>>>>> dropped. The underlying root cause of this is the same for
> both
> >>>>>>>>>>>> stateful ParDo and windowing (essentially, assigning window
> >>>>>>>>>>>> labels is a stateful operation when windowing function is
> merging).
> >>>>>>>>>>>>
> >>>>>>>>>>>> The reason for the current behavior of stateful ParDo seems
> to be
> >>>>>>>>>>>> performance, but is it right to abandon correctness in favor
> of
> >>>>>>>>>>>> performance? Wouldn't it be more consistent to have the
> default
> >>>>>>>>>>>> behavior prefer correctness and when you have the specific
> >>>>>>>>>>>> conditions of state function having special properties, then
> you
> >>>>>>>>>>>> can annotate your DoFn (with something like
> >>>>>>>>>>>> @TimeOrderingAgnostic), which would yield a better
> performance in
> >>>>>>>>>>>> that case?
> >>>>>>>>>>> There are two separable questions here.
> >>>>>>>>>>>
> >>>>>>>>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are
> received
> >>>>>>>>>>> in a specific order? In the current model, it is not. Being
> able to
> >>>>>>>>>>> read, handle, and produced out-of-order data, including late
> data, is
> >>>>>>>>>>> a pretty fundamental property of distributed systems.
> >>>>>>>>>>>
> >>>>>>>>>>> 2) Given that some operations are easier (or possibly only
> possible)
> >>>>>>>>>>> to write when operating on ordered data, and that different
> runners
> >>>>>>>>>>> may have (significantly) cheaper ways to provide this ordering
> than
> >>>>>>>>>>> can be done by the user themselves, should we elevate this to a
> >>>>>>>>>>> property of (Stateful?)DoFns that the runner can provide? I
> think a
> >>>>>>>>>>> compelling argument can be made here that we should.
> >>>>>>>>>>>
> >>>>>>>>>>> - Robert
> >>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
> >>>>>>>>>>>>
> >>>>>>>>>>>> Thanks for the nice small example of a calculation that
> depends
> >>>>>>>>>>>> on order. You are right that many state machines have this
> >>>>>>>>>>>> property. I agree w/ you and Luke that it is convenient for
> batch
> >>>>>>>>>>>> processing to sort by event timestamp before running a
> stateful
> >>>>>>>>>>>> ParDo. In streaming you could also implement "sort by event
> >>>>>>>>>>>> timestamp" by buffering until you know all earlier data will
> be
> >>>>>>>>>>>> dropped - a slack buffer up to allowed lateness.
> >>>>>>>>>>>>
> >>>>>>>>>>>> I do not think that it is OK to sort in batch and not in
> >>>>>>>>>>>> streaming. Many state machines diverge very rapidly when
> things
> >>>>>>>>>>>> are out of order. So each runner if they see the
> >>>>>>>>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
> >>>>>>>>>>>> sorted data (by some mix of buffering and dropping), or to
> reject
> >>>>>>>>>>>> the pipeline as unsupported.
> >>>>>>>>>>>>
> >>>>>>>>>>>> And also want to say that this is not the default case - many
> >>>>>>>>>>>> uses of state & timers in ParDo yield different results at the
> >>>>>>>>>>>> element level, but the results are equivalent at in the big
> >>>>>>>>>>>> picture. Such as the example of "assign a unique sequence
> number
> >>>>>>>>>>>> to each element" or "group into batches" it doesn't matter
> >>>>>>>>>>>> exactly what the result is, only that it meets the spec. And
> >>>>>>>>>>>> other cases like user funnels are monotonic enough that you
> also
> >>>>>>>>>>>> don't actually need sorting.
> >>>>>>>>>>>>
> >>>>>>>>>>>> Kenn
> >>>>>>>>>>>>
> >>>>>>>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je.ik@seznam.cz
> >
> >>>>>>>>>>>> wrote:
> >>>>>>>>>>>>> Yes, the problem will arise probably mostly when you have not
> >>>>>>>>>>>>> well distributed keys (or too few keys). I'm really not sure
> if
> >>>>>>>>>>>>> a pure GBK with a trigger can solve this - it might help to
> have
> >>>>>>>>>>>>> data driven trigger. There would still be some doubts,
> though.
> >>>>>>>>>>>>> The main question is still here - people say, that sorting by
> >>>>>>>>>>>>> timestamp before stateful ParDo would be prohibitively slow,
> but
> >>>>>>>>>>>>> I don't really see why - the sorting is very probably already
> >>>>>>>>>>>>> there. And if not (hash grouping instead of sorted grouping),
> >>>>>>>>>>>>> then the sorting would affect only user defined
> StatefulParDos.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> This would suggest that the best way out of this would be
> really
> >>>>>>>>>>>>> to add annotation, so that the author of the pipeline can
> decide.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> If that would be acceptable I think I can try to prepare some
> >>>>>>>>>>>>> basic functionality, but I'm not sure, if I would be able to
> >>>>>>>>>>>>> cover all runners / sdks.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> It is read all per key and window and not just read all (this
> >>>>>>>>>>>>> still won't scale with hot keys in the global window). The
> GBK
> >>>>>>>>>>>>> preceding the StatefulParDo will guarantee that you are
> >>>>>>>>>>>>> processing all the values for a specific key and window at
> any
> >>>>>>>>>>>>> given time. Is there a specific window/trigger that is
> missing
> >>>>>>>>>>>>> that you feel would remove the need for you to use
> StatefulParDo?
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <
> je.ik@seznam.cz>
> >>>>>>>>>>>>> wrote:
> >>>>>>>>>>>>>> Hi Lukasz,
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Today, if you must have a strict order, you must guarantee
> >>>>>>>>>>>>>>> that your StatefulParDo implements the necessary
> "buffering &
> >>>>>>>>>>>>>>> sorting" into state.
> >>>>>>>>>>>>>> Yes, no problem with that. But this whole discussion
> started,
> >>>>>>>>>>>>>> because *this doesn't work on batch*. You simply cannot
> first
> >>>>>>>>>>>>>> read everything from distributed storage and then buffer it
> all
> >>>>>>>>>>>>>> into memory, just to read it again, but sorted. That will
> not
> >>>>>>>>>>>>>> work. And even if it would, it would be a terrible waste of
> >>>>>>>>>>>>>> resources.
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> Jan
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <
> je.ik@seznam.cz>
> >>>>>>>>>>>>>> wrote:
> >>>>>>>>>>>>>>> This discussion brings many really interesting questions
> for
> >>>>>>>>>>>>>>> me. :-)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > I don't see batch vs. streaming as part of the
> model. One
> >>>>>>>>>>>>>>> can have
> >>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
> different
> >>>>>>>>>>>>>>> modes.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Although I understand motivation of this statement, this
> >>>>>>>>>>>>>>> project name is
> >>>>>>>>>>>>>>> "Apache Beam: An advanced unified programming model". What
> >>>>>>>>>>>>>>> does the
> >>>>>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the
> model?
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Using microbatching, chaining of batch jobs, or pure
> streaming
> >>>>>>>>>>>>>>> are
> >>>>>>>>>>>>>>> exactly the "runtime conditions/characteristics" I refer
> to.
> >>>>>>>>>>>>>>> All these
> >>>>>>>>>>>>>>> define several runtime parameters, which in turn define how
> >>>>>>>>>>>>>>> well/badly
> >>>>>>>>>>>>>>> will the pipeline perform and how many resources might be
> >>>>>>>>>>>>>>> needed. From
> >>>>>>>>>>>>>>> my point of view, pure streaming should be the most
> resource
> >>>>>>>>>>>>>>> demanding
> >>>>>>>>>>>>>>> (if not, why bother with batch? why not run everything in
> >>>>>>>>>>>>>>> streaming
> >>>>>>>>>>>>>>> only? what will there remain to "unify"?).
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > Fortunately, for batch, only the state for a single
> key
> >>>>>>>>>>>>>>> needs to be
> >>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys
> across
> >>>>>>>>>>>>>>> the range
> >>>>>>>>>>>>>>> of skew. Of course if you have few or hot keys, one can
> still
> >>>>>>>>>>>>>>> have
> >>>>>>>>>>>>>>> issues (and this is not specific to StatefulDoFns).
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Yes, but here is still the presumption that my stateful
> DoFn can
> >>>>>>>>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the
> use
> >>>>>>>>>>>>>>> case in
> >>>>>>>>>>>>>>> more detail.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
> >>>>>>>>>>>>>>> some key for
> >>>>>>>>>>>>>>> each element, which is irrelevant for the demonstration).
> Your
> >>>>>>>>>>>>>>> task is
> >>>>>>>>>>>>>>> to calculate in running global window the actual number of
> >>>>>>>>>>>>>>> changes
> >>>>>>>>>>>>>>> between state 0 and state 1 and vice versa. When the state
> >>>>>>>>>>>>>>> doesn't
> >>>>>>>>>>>>>>> change, you don't calculate anything. If input (for given
> key)
> >>>>>>>>>>>>>>> would be
> >>>>>>>>>>>>>>> (tN denotes timestamp N):
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t1: 1
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t2: 0
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t3: 0
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t4: 1
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t5: 1
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t6: 0
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> then the output should yield (supposing that default state
> is
> >>>>>>>>>>>>>>> zero):
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t1: (one: 1, zero: 0)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t2: (one: 1, zero: 1)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t3: (one: 1, zero: 1)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t4: (one: 2, zero: 1)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t5: (one: 2, zero: 1)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> t6: (one: 2, zero: 2)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> How would you implement this in current Beam semantics?
> >>>>>>>>>>>>>> I think your saying here that I know that my input is
> ordered
> >>>>>>>>>>>>>> in a specific way and since I assume the order when writing
> my
> >>>>>>>>>>>>>> pipeline I can perform this optimization. But there is
> nothing
> >>>>>>>>>>>>>> preventing a runner from noticing that your processing in
> the
> >>>>>>>>>>>>>> global window with a specific type of trigger and
> re-ordering
> >>>>>>>>>>>>>> your inputs/processing to get better performance (since you
> >>>>>>>>>>>>>> can't use an AfterWatermark trigger for your pipeline in
> >>>>>>>>>>>>>> streaming for the GlobalWindow).
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> Today, if you must have a strict order, you must guarantee
> that
> >>>>>>>>>>>>>> your StatefulParDo implements the necessary "buffering &
> >>>>>>>>>>>>>> sorting" into state. I can see why you would want an
> annotation
> >>>>>>>>>>>>>> that says I must have timestamp ordered elements, since it
> >>>>>>>>>>>>>> makes writing certain StatefulParDos much easier.
> StatefulParDo
> >>>>>>>>>>>>>> is a low-level function, it really is the "here you go and
> do
> >>>>>>>>>>>>>> whatever you need to but here be dragons" function while
> >>>>>>>>>>>>>> windowing and triggering is meant to keep many people from
> >>>>>>>>>>>>>> writing StatefulParDo in the first place.
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > Pipelines that fail in the "worst case" batch
> scenario
> >>>>>>>>>>>>>>> are likely to
> >>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the
> watermark
> >>>>>>>>>>>>>>> falls
> >>>>>>>>>>>>>>> behind in streaming mode as well.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> But the worst case is defined by input of size (available
> >>>>>>>>>>>>>>> resources +
> >>>>>>>>>>>>>>> single byte) -> pipeline fail. Although it could have
> >>>>>>>>>>>>>>> finished, given
> >>>>>>>>>>>>>>> the right conditions.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > This might be reasonable, implemented by default by
> >>>>>>>>>>>>>>> buffering
> >>>>>>>>>>>>>>> everything and releasing elements as the watermark
> (+lateness)
> >>>>>>>>>>>>>>> advances,
> >>>>>>>>>>>>>>> but would likely lead to inefficient (though *maybe*
> easier to
> >>>>>>>>>>>>>>> reason
> >>>>>>>>>>>>>>> about) code.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Sure, the pipeline will be less efficient, because it would
> >>>>>>>>>>>>>>> have to
> >>>>>>>>>>>>>>> buffer and sort the inputs. But at least it will produce
> >>>>>>>>>>>>>>> correct results
> >>>>>>>>>>>>>>> in cases where updates to state are order-sensitive.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > Would it be roughly equivalent to GBK +
> FlatMap(lambda
> >>>>>>>>>>>>>>> (key, values):
> >>>>>>>>>>>>>>> [(key, value) for value in values])?
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> I'd say roughly yes, but difference would be in the
> trigger.
> >>>>>>>>>>>>>>> The trigger
> >>>>>>>>>>>>>>> should ideally fire as soon as watermark (+lateness)
> crosses
> >>>>>>>>>>>>>>> element
> >>>>>>>>>>>>>>> with lowest timestamp in the buffer. Although this could be
> >>>>>>>>>>>>>>> somehow
> >>>>>>>>>>>>>>> emulated by fixed trigger each X millis.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> > Or is the underlying desire just to be able to hint
> to
> >>>>>>>>>>>>>>> the runner
> >>>>>>>>>>>>>>> that the code may perform better (e.g. require less
> resources)
> >>>>>>>>>>>>>>> as skew
> >>>>>>>>>>>>>>> is reduced (and hence to order by timestamp iff it's
> cheap)?
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> No, the sorting would have to be done in streaming case as
> >>>>>>>>>>>>>>> well. That is
> >>>>>>>>>>>>>>> an imperative of the unified model. I think it is possible
> to
> >>>>>>>>>>>>>>> sort by
> >>>>>>>>>>>>>>> timestamp only in batch case (and do it for *all* batch
> >>>>>>>>>>>>>>> stateful pardos
> >>>>>>>>>>>>>>> without annotation), or introduce annotation, but then make
> >>>>>>>>>>>>>>> the same
> >>>>>>>>>>>>>>> guarantees for streaming case as well.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Jan
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
> >>>>>>>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
> >>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
> >>>>>>>>>>>>>>>>> Hi Robert,
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> yes, I think you rephrased my point - although no
> *explicit*
> >>>>>>>>>>>>>>>>> guarantees
> >>>>>>>>>>>>>>>>> of ordering are given in either mode, there is *implicit*
> >>>>>>>>>>>>>>>>> ordering in
> >>>>>>>>>>>>>>>>> streaming case that is due to nature of the processing -
> the
> >>>>>>>>>>>>>>>>> difference
> >>>>>>>>>>>>>>>>> between watermark and timestamp of elements flowing
> through
> >>>>>>>>>>>>>>>>> the pipeline
> >>>>>>>>>>>>>>>>> are generally low (too high difference leads to the
> >>>>>>>>>>>>>>>>> overbuffering
> >>>>>>>>>>>>>>>>> problem), but there is no such bound on batch.
> >>>>>>>>>>>>>>>> Fortunately, for batch, only the state for a single key
> needs
> >>>>>>>>>>>>>>>> to be
> >>>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys
> >>>>>>>>>>>>>>>> across the
> >>>>>>>>>>>>>>>> range of skew. Of course if you have few or hot keys, one
> can
> >>>>>>>>>>>>>>>> still
> >>>>>>>>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> As a result, I see a few possible solutions:
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> - the best and most natural seems to be extension
> of
> >>>>>>>>>>>>>>>>> the model, so
> >>>>>>>>>>>>>>>>> that it defines batch as not only "streaming pipeline
> >>>>>>>>>>>>>>>>> executed in batch
> >>>>>>>>>>>>>>>>> fashion", but "pipeline with at least as good runtime
> >>>>>>>>>>>>>>>>> characteristics as
> >>>>>>>>>>>>>>>>> in streaming case, executed in batch fashion", I really
> >>>>>>>>>>>>>>>>> don't think that
> >>>>>>>>>>>>>>>>> there are any conflicts with the current model, or that
> this
> >>>>>>>>>>>>>>>>> could
> >>>>>>>>>>>>>>>>> affect performance, because the required sorting (as
> pointed by
> >>>>>>>>>>>>>>>>> Aljoscha) is very probably already done during
> translation
> >>>>>>>>>>>>>>>>> of stateful
> >>>>>>>>>>>>>>>>> pardos. Also note that this definition only affects user
> >>>>>>>>>>>>>>>>> defined
> >>>>>>>>>>>>>>>>> stateful pardos
> >>>>>>>>>>>>>>>> I don't see batch vs. streaming as part of the model. One
> can
> >>>>>>>>>>>>>>>> have
> >>>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
> >>>>>>>>>>>>>>>> different modes.
> >>>>>>>>>>>>>>>> The model describes what the valid outputs are given a
> >>>>>>>>>>>>>>>> (sometimes
> >>>>>>>>>>>>>>>> partial) set of inputs. It becomes really hard to define
> >>>>>>>>>>>>>>>> things like
> >>>>>>>>>>>>>>>> "as good runtime characteristics." Once you allow any
> >>>>>>>>>>>>>>>> out-of-orderedness, it is not very feasible to try and
> define
> >>>>>>>>>>>>>>>> (and
> >>>>>>>>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
> >>>>>>>>>>>>>>>> out-of-orderedness.
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
> >>>>>>>>>>>>>>>> likely to
> >>>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the
> watermark
> >>>>>>>>>>>>>>>> falls
> >>>>>>>>>>>>>>>> behind in streaming mode as well.
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> - another option would be to introduce annotation
> for
> >>>>>>>>>>>>>>>>> DoFns (e.g.
> >>>>>>>>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result
> in
> >>>>>>>>>>>>>>>>> the sorting
> >>>>>>>>>>>>>>>>> in batch case - but - this extension would have to ensure
> >>>>>>>>>>>>>>>>> the sorting in
> >>>>>>>>>>>>>>>>> streaming mode also - it would require definition of
> allowed
> >>>>>>>>>>>>>>>>> lateness,
> >>>>>>>>>>>>>>>>> and triggger (essentially similar to window)
> >>>>>>>>>>>>>>>> This might be reasonable, implemented by default by
> buffering
> >>>>>>>>>>>>>>>> everything and releasing elements as the watermark
> (+lateness)
> >>>>>>>>>>>>>>>> advances, but would likely lead to inefficient (though
> >>>>>>>>>>>>>>>> *maybe* easier
> >>>>>>>>>>>>>>>> to reason about) code. Not sure about the semantics of
> >>>>>>>>>>>>>>>> triggering
> >>>>>>>>>>>>>>>> here, especially data-driven triggers. Would it be roughly
> >>>>>>>>>>>>>>>> equivalent
> >>>>>>>>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
> >>>>>>>>>>>>>>>> value in
> >>>>>>>>>>>>>>>> values])?
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> Or is the underlying desire just to be able to hint to the
> >>>>>>>>>>>>>>>> runner that
> >>>>>>>>>>>>>>>> the code may perform better (e.g. require less resources)
> as
> >>>>>>>>>>>>>>>> skew is
> >>>>>>>>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> - last option would be to introduce these "higher
> order
> >>>>>>>>>>>>>>>>> guarantees" in
> >>>>>>>>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be
> the
> >>>>>>>>>>>>>>>>> worst
> >>>>>>>>>>>>>>>>> option to me
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> I see the first two options quite equally good, although
> the
> >>>>>>>>>>>>>>>>> letter one
> >>>>>>>>>>>>>>>>> is probably more time consuming to implement. But it
> would
> >>>>>>>>>>>>>>>>> bring
> >>>>>>>>>>>>>>>>> additional feature to streaming case as well.
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> Thanks for any thoughts.
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> Jan
> >>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
> >>>>>>>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
> >>>>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
> >>>>>>>>>>>>>>>>>>> Hi Reuven,
> >>>>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
> >>>>>>>>>>>>>>>>>>>> runners.
> >>>>>>>>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic
> inside
> >>>>>>>>>>>>>>>>>>> the state works for absolutely unbounded
> out-of-orderness
> >>>>>>>>>>>>>>>>>>> of elements. That basically (practically) can work only
> >>>>>>>>>>>>>>>>>>> for cases, where the order of input elements doesn't
> >>>>>>>>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and
> any
> >>>>>>>>>>>>>>>>>>> time you have a state machine involved, then the
> ordering
> >>>>>>>>>>>>>>>>>>> of elements would matter.
> >>>>>>>>>>>>>>>>>> No guarantees on order are provided in *either*
> streaming
> >>>>>>>>>>>>>>>>>> or batch
> >>>>>>>>>>>>>>>>>> mode by the model. However, it is the case that in
> order to
> >>>>>>>>>>>>>>>>>> make
> >>>>>>>>>>>>>>>>>> forward progress most streaming runners attempt to limit
> >>>>>>>>>>>>>>>>>> the amount of
> >>>>>>>>>>>>>>>>>> out-of-orderedness of elements (in terms of event time
> vs.
> >>>>>>>>>>>>>>>>>> processing
> >>>>>>>>>>>>>>>>>> time) to make forward progress, which in turn could help
> >>>>>>>>>>>>>>>>>> cap the
> >>>>>>>>>>>>>>>>>> amount of state that must be held concurrently, whereas
> a
> >>>>>>>>>>>>>>>>>> batch runner
> >>>>>>>>>>>>>>>>>> may not allow any state to be safely discarded until
> the whole
> >>>>>>>>>>>>>>>>>> timeline from infinite past to infinite future has been
> >>>>>>>>>>>>>>>>>> observed.
> >>>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
> >>>>>>>>>>>>>>>>>> batch" in batch mode.
> >>>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
> >>>>>>>>>>>>>>>>>> <mx...@apache.org> wrote:
> >>>>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>>>>> batch semantics and streaming semantics differs
> only
> >>>>>>>>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger
> on
> >>>>>>>>>>>>>>>>>>>> batch and cannot on stream
> >>>>>>>>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
> >>>>>>>>>>>>>>>>>>> trigger. You
> >>>>>>>>>>>>>>>>>>> could define additional triggers that do early firings.
> >>>>>>>>>>>>>>>>>>> And you could
> >>>>>>>>>>>>>>>>>>> even trigger the global window by advancing the
> watermark
> >>>>>>>>>>>>>>>>>>> to +inf.
> >>>>>>>>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window
> with
> >>>>>>>>>>>>>>>>>> default
> >>>>>>>>>>>>>>>>>> trigger on unbounded PCollections in the SDK because
> this
> >>>>>>>>>>>>>>>>>> is more
> >>>>>>>>>>>>>>>>>> likely to be user error than an actual desire to have no
> >>>>>>>>>>>>>>>>>> output until
> >>>>>>>>>>>>>>>>>> drain. But it's semantically valid in the model.
>
>
Re: Definition of Unified model
Posted by Jan Lukavský <je...@seznam.cz>.
As I understood it, Kenn was supporting the idea that sequence metadata
is preferable over FIFO. I was trying to point out, that it even should
provide the same functionally as FIFO, plus one important more -
reproducibility and ability to being persisted and reused the same way
in batch and streaming.
There is no doubt, that sequence metadata can be stored in every
storage. But, regarding some implicit ordering that sources might have -
yes, of course, data written into HDFS or Cloud Storage has ordering,
but only partial - inside some bulk (e.g. file) and the ordering is not
defined correctly on boundaries of these bulks (between files). That is
why I'd say, that ordering of sources is relevant only for
(partitioned!) streaming sources and generally always reduces to
sequence metadata (e.g. offsets).
Jan
On 5/28/19 11:43 AM, Robert Bradshaw wrote:
> Huge +1 to all Kenn said.
>
> Jan, batch sources can have orderings too, just like Kafka. I think
> it's reasonable (for both batch and streaming) that if a source has an
> ordering that is an important part of the data, it should preserve
> this ordering into the data itself (e.g. as sequence numbers, offsets,
> etc.)
>
> On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org> wrote:
>> I strongly prefer explicit sequence metadata over FIFO requirements, because:
>>
>> - FIFO is complex to specify: for example Dataflow has "per stage key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage" is not a portable concept, nor even guaranteed to remain a Dataflow concept)
>> - complex specifications are by definition poor usability (if necessary, then it is what it is)
>> - overly restricts the runner, reduces parallelism, for example any non-stateful ParDo has per-element parallelism, not per "key"
>> - another perspective on that: FIFO makes everyone pay rather than just the transform that requires exactly sequencing
>> - previous implementation details like reshuffles become part of the model
>> - I'm not even convinced the use cases involved are addressed by some careful FIFO restrictions; many sinks re-key and they would all have to become aware of how keying of a sequence of "stages" affects the end-to-end FIFO
>>
>> A noop becoming a non-noop is essentially the mathematical definition of moving from higher-level to lower-level abstraction.
>>
>> So this strikes at the core question of what level of abstraction Beam aims to represent. Lower-level means there are fewer possible implementations and it is more tied to the underlying architecture, and anything not near-exact match pays a huge penalty. Higher-level means there are more implementations possible with different tradeoffs, though they may all pay a minor penalty.
>>
>> I could be convinced to change my mind, but it needs some extensive design, examples, etc. I think it is probably about the most consequential design decision in the whole Beam model, around the same level as the decision to use ParDo and GBK as the primitives IMO.
>>
>> Kenn
>>
>> On Thu, May 23, 2019 at 10:17 AM Reuven Lax <re...@google.com> wrote:
>>> Not really. I'm suggesting that some variant of FIFO ordering is necessary, which requires either runners natively support FIFO ordering or transforms adding some extra sequence number to each record to sort by.
>>>
>>> I still think your proposal is very useful by the way. I'm merely pointing out that to solve the state-machine problem we probably need something more.
>>>
>>> Reuven
>>>
>>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>> Hi,
>>>> yes. It seems that ordering by user supplied UDF makes sense and I will update the design proposal accordingly.
>>>> Would that solve the issues you mention?
>>>> Jan
>>>> ---------- Původní e-mail ----------
>>>> Od: Reuven Lax <re...@google.com>
>>>> Komu: dev <de...@beam.apache.org>
>>>> Datum: 23. 5. 2019 18:44:38
>>>> Předmět: Re: Definition of Unified model
>>>>
>>>> I'm simply saying that timestamp ordering is insufficient for state machines. I wasn't proposing Kafka as a solution - that was simply an example of how people solve this problem in other scenarios.
>>>>
>>>> BTW another example of ordering: Imagine today that you have a triggered Sum aggregation writing out to a key-value sink. In theory we provide no ordering, so the sink might write the triggered sums in the wrong order, ending up with an incorrect value in the sink. In this case you probably want values ordered by trigger pane index.
>>>>
>>>> Reuven
>>>>
>>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>>
>>>> Hi Reuven,
>>>> I share the view point of Robert. I think the isuue you refer to is not in reality related to timestamps, but to the fact, that ordering of events in time is observer dependent (either caused by relativity, or time skew, essentially this has the same consequences). And the resolution in fact isn't Kafka, but generally an authoritative observer, that tells you "I saw the events in this order". And you either have one (and have the outcome of his observation persisted in the data - e.g. as offset in Kafka partition), then you should be able to use it (maybe that suggests afterall that sorting by some user supplied UDF might make sense), or do not have it, and then any interpretation of the data seems to be equally valid. Although determinism is fine, of course.
>>>> Jan
>>>> ---------- Původní e-mail ----------
>>>> Od: Reuven Lax <re...@google.com>
>>>> Komu: dev <de...@beam.apache.org>
>>>> Datum: 23. 5. 2019 17:39:12
>>>> Předmět: Re: Definition of Unified model
>>>>
>>>> So an example would be elements of type "startUserSession" and "endUserSession" (website sessions, not Beam sessions). Logically you may need to process them in the correct order if you have any sort of state-machine logic. However timestamp ordering is never guaranteed to match the logical ordering. Not only might you have several elements with the same timestamp, but in reality time skew across backend servers can cause the events to have timestamps in reverse order of the actual causality order.
>>>>
>>>> People do solve this problem today though. Publish the events to Kafka, making sure that events for the same user end up in the same Kafka partition. This ensures that the events appear in the Kafka partitions in causality order, even if the timestamp order doesn't match. The your Kafka subscriber simply process the elements in each partition in order.
>>>>
>>>> I think the ability to impose FIFO causality ordering is what's needed for any state-machine work. Timestamp ordering has advantages (though often I think the advantage is in state), but does not solve this problem.
>>>>
>>>> Reuven
>>>>
>>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <ro...@google.com> wrote:
>>>>
>>>> Good point.
>>>>
>>>> The "implementation-specific" way I would do this is
>>>> window-by-instant, followed by a DoFn that gets all the elements with
>>>> the same timestamp and sorts/acts accordingly, but this counts on the
>>>> runner producing windows in timestamp order (likely?) and also the
>>>> subsequent DoFn getting them in this order (also likely, due to
>>>> fusion).
>>>>
>>>> One could make the argument that, though it does not provide
>>>> deterministic behavior, getting elements of the same timestamp in
>>>> different orders should produce equally valid interpretations of the
>>>> data. (After all, due to relatively, timestamps are not technically
>>>> well ordered across space.) I can see how data-dependent tiebreakers
>>>> could be useful, or promises of preservation of order between
>>>> operations.
>>>>
>>>> - Robert
>>>>
>>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com> wrote:
>>>>> So Jan's example of state machines is quite a valid use case for ordering. However in my experience, timestamp ordering is insufficient for state machines. Elements that cause state transitions might come in with the exact same timestamp, yet still have a necessary ordering. Especially given Beam's decision to have milliseconds timestamps this is possible, but even at microsecond or nanosecond precision this can happen at scale. To handle state machines you usually need some sort of FIFO ordering along with an ordered sources, such as Kafka, not timestamp ordering.
>>>>>
>>>>> Reuven
>>>>>
>>>>> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>>>> Hi all,
>>>>>>
>>>>>> thanks everyone for this discussion. I think I have gathered enough
>>>>>> feedback to be able to put down a proposition for changes, which I will
>>>>>> do and send to this list for further discussion. There are still doubts
>>>>>> remaining the non-determinism and it's relation to outputs stability vs.
>>>>>> latency. But I will try to clarify all this in the design document.
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> Jan
>>>>>>
>>>>>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
>>>>>>>> current understanding.
>>>>>>> In essence your description of how exactly-once works in Flink is
>>>>>>> correct. The general assumption in Flink is that pipelines must be
>>>>>>> deterministic and thus produce idempotent writes in the case of
>>>>>>> failures. However, that doesn't mean Beam sinks can't guarantee a bit
>>>>>>> more with what Flink has to offer.
>>>>>>>
>>>>>>> Luke already mentioned the design discussions for @RequiresStableInput
>>>>>>> which ensures idempotent writes for non-deterministic pipelines. This
>>>>>>> is not part of the model but an optional Beam feature.
>>>>>>>
>>>>>>> We recently implemented support for @RequiresStableInput in the Flink
>>>>>>> Runner. Reuven mentioned the Flink checkpoint confirmation, which
>>>>>>> allows us to buffer (and checkpoint) processed data and only emit it
>>>>>>> once a Flink checkpoint has completed.
>>>>>>>
>>>>>>> Cheers,
>>>>>>> Max
>>>>>>>
>>>>>>> On 21.05.19 16:49, Jan Lukavský wrote:
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> > Actually, I think it is a larger (open) question whether exactly
>>>>>>>> once is guaranteed by the model or whether runners are allowed to
>>>>>>>> relax that. I would think, however, that sources correctly
>>>>>>>> implemented should be idempotent when run atop an exactly once
>>>>>>>> infrastructure such as Flink of Dataflow.
>>>>>>>>
>>>>>>>> I would assume, that the model basically inherits guarantees of
>>>>>>>> underlying infrastructure. Because Flink does not work as you
>>>>>>>> described (atomic commit of inputs, state and outputs), but rather a
>>>>>>>> checkpoint mark is flowing through the DAG much like watermark and on
>>>>>>>> failures operators are restored and data reprocessed, it (IMHO)
>>>>>>>> implies, that you have exactly once everywhere in the DAG *but*
>>>>>>>> sinks. That is because sinks cannot be restored to previous state,
>>>>>>>> instead sinks are supposed to be idempotent in order for the exactly
>>>>>>>> once to really work (or at least be able to commit outputs on
>>>>>>>> checkpoint in sink). That implies that if you don't have sink that is
>>>>>>>> able to commit outputs atomically on checkpoint, the pipeline
>>>>>>>> execution should be deterministic upon retries, otherwise shadow
>>>>>>>> writes from failed paths of the pipeline might appear.
>>>>>>>>
>>>>>>>> Someone from Flink might correct me if I'm wrong, but that's my
>>>>>>>> current understanding.
>>>>>>>>
>>>>>>>> > Sounds like we should make this clearer.
>>>>>>>>
>>>>>>>> I meant that you are right that we must not in any thoughts we are
>>>>>>>> having forget that streams are by definition out-of-order. That is
>>>>>>>> property that we cannot change. But - that doesn't limit us from
>>>>>>>> creating operator that presents the data to UDF as if the stream was
>>>>>>>> ideally sorted. It can do that by introducing latency, of course.
>>>>>>>>
>>>>>>>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>>>>>>>>> Reza: One could provide something like this as a utility class, but
>>>>>>>>> one downside is that it is not scale invariant. It requires a tuning
>>>>>>>>> parameter that, if to small, won't mitigate the problem, but if to
>>>>>>>>> big, greatly increases latency. (Possibly one could define a dynamic
>>>>>>>>> session-like window to solve this though...) It also might be harder
>>>>>>>>> for runners that *can* cheaply present stuff in timestamp order to
>>>>>>>>> optimize. (That and, in practice, our annotation-style process methods
>>>>>>>>> don't lend themselves to easy composition.) I think it could work in
>>>>>>>>> specific cases though.
>>>>>>>>>
>>>>>>>>> More inline below.
>>>>>>>>>
>>>>>>>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>>>>>>>> Hi Robert,
>>>>>>>>>>
>>>>>>>>>> > Beam has an exactly-once model. If the data was consumed, state
>>>>>>>>>> mutated, and outputs written downstream (these three are committed
>>>>>>>>>> together atomically) it will not be replayed. That does not, of
>>>>>>>>>> course,
>>>>>>>>>> solve the non-determanism due to ordering (including the fact that two
>>>>>>>>>> operations reading the same PCollection may view different ordering).
>>>>>>>>>>
>>>>>>>>>> I think what you describe is a property of a runner, not of the model,
>>>>>>>>>> right? I think if I run my pipeline on Flink I will not get this
>>>>>>>>>> atomicity, because although Flink uses also exactly-once model if
>>>>>>>>>> might
>>>>>>>>>> write outputs multiple times.
>>>>>>>>> Actually, I think it is a larger (open) question whether exactly once
>>>>>>>>> is guaranteed by the model or whether runners are allowed to relax
>>>>>>>>> that. I would think, however, that sources correctly implemented
>>>>>>>>> should be idempotent when run atop an exactly once infrastructure such
>>>>>>>>> as Flink of Dataflow.
>>>>>>>>>
>>>>>>>>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
>>>>>>>>>> received
>>>>>>>>>> in a specific order? In the current model, it is not. Being able to
>>>>>>>>>> read, handle, and produced out-of-order data, including late data,
>>>>>>>>>> is a
>>>>>>>>>> pretty fundamental property of distributed systems.
>>>>>>>>>>
>>>>>>>>>> Yes, absolutely. The argument here is not that Stateful ParDo should
>>>>>>>>>> presume to receive elements in any order, but to _present_ it as
>>>>>>>>>> such to
>>>>>>>>>> the user @ProcessElement function.
>>>>>>>>> Sounds like we should make this clearer.
>>>>>>>>>
>>>>>>>>>> > 2) Given that some operations are easier (or possibly only
>>>>>>>>>> possible)
>>>>>>>>>> to write when operating on ordered data, and that different runners
>>>>>>>>>> may
>>>>>>>>>> have (significantly) cheaper ways to provide this ordering than can be
>>>>>>>>>> done by the user themselves, should we elevate this to a property of
>>>>>>>>>> (Stateful?)DoFns that the runner can provide? I think a compelling
>>>>>>>>>> argument can be made here that we should.
>>>>>>>>>>
>>>>>>>>>> +1
>>>>>>>>>>
>>>>>>>>>> Jan
>>>>>>>>>>
>>>>>>>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>>>>>>>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz> wrote:
>>>>>>>>>>>> > I don't see batch vs. streaming as part of the model. One
>>>>>>>>>>>> can have
>>>>>>>>>>>> microbatch, or even a runner that alternates between different
>>>>>>>>>>>> modes.
>>>>>>>>>>>>
>>>>>>>>>>>> Although I understand motivation of this statement, this project
>>>>>>>>>>>> name is
>>>>>>>>>>>> "Apache Beam: An advanced unified programming model". What does the
>>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>>>>>>>>>>> What I mean is that streaming vs. batch is no longer part of the
>>>>>>>>>>> model
>>>>>>>>>>> (or ideally API), but pushed down to be a concern of the runner
>>>>>>>>>>> (executor) of the pipeline.
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz>
>>>>>>>>>>> wrote:
>>>>>>>>>>>> Hi Kenn,
>>>>>>>>>>>>
>>>>>>>>>>>> OK, so if we introduce annotation, we can have stateful ParDo
>>>>>>>>>>>> with sorting, that would perfectly resolve my issues. I still
>>>>>>>>>>>> have some doubts, though. Let me explain. The current behavior of
>>>>>>>>>>>> stateful ParDo has the following properties:
>>>>>>>>>>>>
>>>>>>>>>>>> a) might fail in batch, although runs fine in streaming (that
>>>>>>>>>>>> is due to the buffering, and unbounded lateness in batch, which
>>>>>>>>>>>> was discussed back and forth in this thread)
>>>>>>>>>>>>
>>>>>>>>>>>> b) might be non deterministic (this is because the elements
>>>>>>>>>>>> arrive at somewhat random order, and even if you do the operation
>>>>>>>>>>>> "assign unique ID to elements" this might produce different
>>>>>>>>>>>> results when run multiple times)
>>>>>>>>>>> PCollections are *explicitly* unordered. Any operations that
>>>>>>>>>>> assume or
>>>>>>>>>>> depend on a specific ordering for correctness (or determinism) must
>>>>>>>>>>> provide that ordering themselves (i.e. tolerate "arbitrary shuffling
>>>>>>>>>>> of inputs"). As you point out, that may be very expensive if you have
>>>>>>>>>>> very hot keys with very large (unbounded) timestamp skew.
>>>>>>>>>>>
>>>>>>>>>>> StatefulDoFns are low-level operations that should be used with care;
>>>>>>>>>>> the simpler windowing model gives determinism in the face of
>>>>>>>>>>> unordered
>>>>>>>>>>> data (though late data and non-end-of-window triggering introduces
>>>>>>>>>>> some of the non-determanism back in).
>>>>>>>>>>>
>>>>>>>>>>>> What worries me most is the property b), because it seems to me
>>>>>>>>>>>> to have serious consequences - not only that if you run twice
>>>>>>>>>>>> batch pipeline you would get different results, but even on
>>>>>>>>>>>> streaming, when pipeline fails and gets restarted from
>>>>>>>>>>>> checkpoint, produced output might differ from the previous run
>>>>>>>>>>>> and data from the first run might have already been persisted
>>>>>>>>>>>> into sink. That would create somewhat messy outputs.
>>>>>>>>>>> Beam has an exactly-once model. If the data was consumed, state
>>>>>>>>>>> mutated, and outputs written downstream (these three are committed
>>>>>>>>>>> together atomically) it will not be replayed. That does not, of
>>>>>>>>>>> course, solve the non-determanism due to ordering (including the fact
>>>>>>>>>>> that two operations reading the same PCollection may view different
>>>>>>>>>>> ordering).
>>>>>>>>>>>
>>>>>>>>>>>> These two properties makes me think that the current
>>>>>>>>>>>> implementation is more of a _special case_ than the general one.
>>>>>>>>>>>> The general one would be that your state doesn't have the
>>>>>>>>>>>> properties to be able to tolerate buffering problems and/or
>>>>>>>>>>>> non-determinism. Which is the case where you need sorting in both
>>>>>>>>>>>> streaming and batch to be part of the model.
>>>>>>>>>>>>
>>>>>>>>>>>> Let me point out one more analogy - that is merging vs.
>>>>>>>>>>>> non-merging windows. The general case (merging windows) implies
>>>>>>>>>>>> sorting by timestamp in both batch case (explicit) and streaming
>>>>>>>>>>>> (buffering). The special case (non-merging windows) doesn't rely
>>>>>>>>>>>> on any timestamp ordering, so the sorting and buffering can be
>>>>>>>>>>>> dropped. The underlying root cause of this is the same for both
>>>>>>>>>>>> stateful ParDo and windowing (essentially, assigning window
>>>>>>>>>>>> labels is a stateful operation when windowing function is merging).
>>>>>>>>>>>>
>>>>>>>>>>>> The reason for the current behavior of stateful ParDo seems to be
>>>>>>>>>>>> performance, but is it right to abandon correctness in favor of
>>>>>>>>>>>> performance? Wouldn't it be more consistent to have the default
>>>>>>>>>>>> behavior prefer correctness and when you have the specific
>>>>>>>>>>>> conditions of state function having special properties, then you
>>>>>>>>>>>> can annotate your DoFn (with something like
>>>>>>>>>>>> @TimeOrderingAgnostic), which would yield a better performance in
>>>>>>>>>>>> that case?
>>>>>>>>>>> There are two separable questions here.
>>>>>>>>>>>
>>>>>>>>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are received
>>>>>>>>>>> in a specific order? In the current model, it is not. Being able to
>>>>>>>>>>> read, handle, and produced out-of-order data, including late data, is
>>>>>>>>>>> a pretty fundamental property of distributed systems.
>>>>>>>>>>>
>>>>>>>>>>> 2) Given that some operations are easier (or possibly only possible)
>>>>>>>>>>> to write when operating on ordered data, and that different runners
>>>>>>>>>>> may have (significantly) cheaper ways to provide this ordering than
>>>>>>>>>>> can be done by the user themselves, should we elevate this to a
>>>>>>>>>>> property of (Stateful?)DoFns that the runner can provide? I think a
>>>>>>>>>>> compelling argument can be made here that we should.
>>>>>>>>>>>
>>>>>>>>>>> - Robert
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>>>>>>>>>>>>
>>>>>>>>>>>> Thanks for the nice small example of a calculation that depends
>>>>>>>>>>>> on order. You are right that many state machines have this
>>>>>>>>>>>> property. I agree w/ you and Luke that it is convenient for batch
>>>>>>>>>>>> processing to sort by event timestamp before running a stateful
>>>>>>>>>>>> ParDo. In streaming you could also implement "sort by event
>>>>>>>>>>>> timestamp" by buffering until you know all earlier data will be
>>>>>>>>>>>> dropped - a slack buffer up to allowed lateness.
>>>>>>>>>>>>
>>>>>>>>>>>> I do not think that it is OK to sort in batch and not in
>>>>>>>>>>>> streaming. Many state machines diverge very rapidly when things
>>>>>>>>>>>> are out of order. So each runner if they see the
>>>>>>>>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
>>>>>>>>>>>> sorted data (by some mix of buffering and dropping), or to reject
>>>>>>>>>>>> the pipeline as unsupported.
>>>>>>>>>>>>
>>>>>>>>>>>> And also want to say that this is not the default case - many
>>>>>>>>>>>> uses of state & timers in ParDo yield different results at the
>>>>>>>>>>>> element level, but the results are equivalent at in the big
>>>>>>>>>>>> picture. Such as the example of "assign a unique sequence number
>>>>>>>>>>>> to each element" or "group into batches" it doesn't matter
>>>>>>>>>>>> exactly what the result is, only that it meets the spec. And
>>>>>>>>>>>> other cases like user funnels are monotonic enough that you also
>>>>>>>>>>>> don't actually need sorting.
>>>>>>>>>>>>
>>>>>>>>>>>> Kenn
>>>>>>>>>>>>
>>>>>>>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz>
>>>>>>>>>>>> wrote:
>>>>>>>>>>>>> Yes, the problem will arise probably mostly when you have not
>>>>>>>>>>>>> well distributed keys (or too few keys). I'm really not sure if
>>>>>>>>>>>>> a pure GBK with a trigger can solve this - it might help to have
>>>>>>>>>>>>> data driven trigger. There would still be some doubts, though.
>>>>>>>>>>>>> The main question is still here - people say, that sorting by
>>>>>>>>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
>>>>>>>>>>>>> I don't really see why - the sorting is very probably already
>>>>>>>>>>>>> there. And if not (hash grouping instead of sorted grouping),
>>>>>>>>>>>>> then the sorting would affect only user defined StatefulParDos.
>>>>>>>>>>>>>
>>>>>>>>>>>>> This would suggest that the best way out of this would be really
>>>>>>>>>>>>> to add annotation, so that the author of the pipeline can decide.
>>>>>>>>>>>>>
>>>>>>>>>>>>> If that would be acceptable I think I can try to prepare some
>>>>>>>>>>>>> basic functionality, but I'm not sure, if I would be able to
>>>>>>>>>>>>> cover all runners / sdks.
>>>>>>>>>>>>>
>>>>>>>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>>>>>>>>>>>>>
>>>>>>>>>>>>> It is read all per key and window and not just read all (this
>>>>>>>>>>>>> still won't scale with hot keys in the global window). The GBK
>>>>>>>>>>>>> preceding the StatefulParDo will guarantee that you are
>>>>>>>>>>>>> processing all the values for a specific key and window at any
>>>>>>>>>>>>> given time. Is there a specific window/trigger that is missing
>>>>>>>>>>>>> that you feel would remove the need for you to use StatefulParDo?
>>>>>>>>>>>>>
>>>>>>>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
>>>>>>>>>>>>> wrote:
>>>>>>>>>>>>>> Hi Lukasz,
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Today, if you must have a strict order, you must guarantee
>>>>>>>>>>>>>>> that your StatefulParDo implements the necessary "buffering &
>>>>>>>>>>>>>>> sorting" into state.
>>>>>>>>>>>>>> Yes, no problem with that. But this whole discussion started,
>>>>>>>>>>>>>> because *this doesn't work on batch*. You simply cannot first
>>>>>>>>>>>>>> read everything from distributed storage and then buffer it all
>>>>>>>>>>>>>> into memory, just to read it again, but sorted. That will not
>>>>>>>>>>>>>> work. And even if it would, it would be a terrible waste of
>>>>>>>>>>>>>> resources.
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Jan
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
>>>>>>>>>>>>>> wrote:
>>>>>>>>>>>>>>> This discussion brings many really interesting questions for
>>>>>>>>>>>>>>> me. :-)
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> > I don't see batch vs. streaming as part of the model. One
>>>>>>>>>>>>>>> can have
>>>>>>>>>>>>>>> microbatch, or even a runner that alternates between different
>>>>>>>>>>>>>>> modes.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Although I understand motivation of this statement, this
>>>>>>>>>>>>>>> project name is
>>>>>>>>>>>>>>> "Apache Beam: An advanced unified programming model". What
>>>>>>>>>>>>>>> does the
>>>>>>>>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
>>>>>>>>>>>>>>> are
>>>>>>>>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
>>>>>>>>>>>>>>> All these
>>>>>>>>>>>>>>> define several runtime parameters, which in turn define how
>>>>>>>>>>>>>>> well/badly
>>>>>>>>>>>>>>> will the pipeline perform and how many resources might be
>>>>>>>>>>>>>>> needed. From
>>>>>>>>>>>>>>> my point of view, pure streaming should be the most resource
>>>>>>>>>>>>>>> demanding
>>>>>>>>>>>>>>> (if not, why bother with batch? why not run everything in
>>>>>>>>>>>>>>> streaming
>>>>>>>>>>>>>>> only? what will there remain to "unify"?).
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> > Fortunately, for batch, only the state for a single key
>>>>>>>>>>>>>>> needs to be
>>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys across
>>>>>>>>>>>>>>> the range
>>>>>>>>>>>>>>> of skew. Of course if you have few or hot keys, one can still
>>>>>>>>>>>>>>> have
>>>>>>>>>>>>>>> issues (and this is not specific to StatefulDoFns).
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Yes, but here is still the presumption that my stateful DoFn can
>>>>>>>>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
>>>>>>>>>>>>>>> case in
>>>>>>>>>>>>>>> more detail.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
>>>>>>>>>>>>>>> some key for
>>>>>>>>>>>>>>> each element, which is irrelevant for the demonstration). Your
>>>>>>>>>>>>>>> task is
>>>>>>>>>>>>>>> to calculate in running global window the actual number of
>>>>>>>>>>>>>>> changes
>>>>>>>>>>>>>>> between state 0 and state 1 and vice versa. When the state
>>>>>>>>>>>>>>> doesn't
>>>>>>>>>>>>>>> change, you don't calculate anything. If input (for given key)
>>>>>>>>>>>>>>> would be
>>>>>>>>>>>>>>> (tN denotes timestamp N):
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t1: 1
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t2: 0
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t3: 0
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t4: 1
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t5: 1
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t6: 0
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> then the output should yield (supposing that default state is
>>>>>>>>>>>>>>> zero):
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t1: (one: 1, zero: 0)
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t2: (one: 1, zero: 1)
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t3: (one: 1, zero: 1)
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t4: (one: 2, zero: 1)
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t5: (one: 2, zero: 1)
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> t6: (one: 2, zero: 2)
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> How would you implement this in current Beam semantics?
>>>>>>>>>>>>>> I think your saying here that I know that my input is ordered
>>>>>>>>>>>>>> in a specific way and since I assume the order when writing my
>>>>>>>>>>>>>> pipeline I can perform this optimization. But there is nothing
>>>>>>>>>>>>>> preventing a runner from noticing that your processing in the
>>>>>>>>>>>>>> global window with a specific type of trigger and re-ordering
>>>>>>>>>>>>>> your inputs/processing to get better performance (since you
>>>>>>>>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>>>>>>>>>>>>>> streaming for the GlobalWindow).
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Today, if you must have a strict order, you must guarantee that
>>>>>>>>>>>>>> your StatefulParDo implements the necessary "buffering &
>>>>>>>>>>>>>> sorting" into state. I can see why you would want an annotation
>>>>>>>>>>>>>> that says I must have timestamp ordered elements, since it
>>>>>>>>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
>>>>>>>>>>>>>> is a low-level function, it really is the "here you go and do
>>>>>>>>>>>>>> whatever you need to but here be dragons" function while
>>>>>>>>>>>>>> windowing and triggering is meant to keep many people from
>>>>>>>>>>>>>> writing StatefulParDo in the first place.
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
>>>>>>>>>>>>>>> are likely to
>>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>>>>>>>>>>>>>>> falls
>>>>>>>>>>>>>>> behind in streaming mode as well.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> But the worst case is defined by input of size (available
>>>>>>>>>>>>>>> resources +
>>>>>>>>>>>>>>> single byte) -> pipeline fail. Although it could have
>>>>>>>>>>>>>>> finished, given
>>>>>>>>>>>>>>> the right conditions.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> > This might be reasonable, implemented by default by
>>>>>>>>>>>>>>> buffering
>>>>>>>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>>>>>>>>>>>>>>> advances,
>>>>>>>>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
>>>>>>>>>>>>>>> reason
>>>>>>>>>>>>>>> about) code.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Sure, the pipeline will be less efficient, because it would
>>>>>>>>>>>>>>> have to
>>>>>>>>>>>>>>> buffer and sort the inputs. But at least it will produce
>>>>>>>>>>>>>>> correct results
>>>>>>>>>>>>>>> in cases where updates to state are order-sensitive.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
>>>>>>>>>>>>>>> (key, values):
>>>>>>>>>>>>>>> [(key, value) for value in values])?
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
>>>>>>>>>>>>>>> The trigger
>>>>>>>>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
>>>>>>>>>>>>>>> element
>>>>>>>>>>>>>>> with lowest timestamp in the buffer. Although this could be
>>>>>>>>>>>>>>> somehow
>>>>>>>>>>>>>>> emulated by fixed trigger each X millis.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> > Or is the underlying desire just to be able to hint to
>>>>>>>>>>>>>>> the runner
>>>>>>>>>>>>>>> that the code may perform better (e.g. require less resources)
>>>>>>>>>>>>>>> as skew
>>>>>>>>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> No, the sorting would have to be done in streaming case as
>>>>>>>>>>>>>>> well. That is
>>>>>>>>>>>>>>> an imperative of the unified model. I think it is possible to
>>>>>>>>>>>>>>> sort by
>>>>>>>>>>>>>>> timestamp only in batch case (and do it for *all* batch
>>>>>>>>>>>>>>> stateful pardos
>>>>>>>>>>>>>>> without annotation), or introduce annotation, but then make
>>>>>>>>>>>>>>> the same
>>>>>>>>>>>>>>> guarantees for streaming case as well.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Jan
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>>>>>>>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>>>>>>>>>>>>>>>>> Hi Robert,
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
>>>>>>>>>>>>>>>>> guarantees
>>>>>>>>>>>>>>>>> of ordering are given in either mode, there is *implicit*
>>>>>>>>>>>>>>>>> ordering in
>>>>>>>>>>>>>>>>> streaming case that is due to nature of the processing - the
>>>>>>>>>>>>>>>>> difference
>>>>>>>>>>>>>>>>> between watermark and timestamp of elements flowing through
>>>>>>>>>>>>>>>>> the pipeline
>>>>>>>>>>>>>>>>> are generally low (too high difference leads to the
>>>>>>>>>>>>>>>>> overbuffering
>>>>>>>>>>>>>>>>> problem), but there is no such bound on batch.
>>>>>>>>>>>>>>>> Fortunately, for batch, only the state for a single key needs
>>>>>>>>>>>>>>>> to be
>>>>>>>>>>>>>>>> preserved at a time, rather than the state for all keys
>>>>>>>>>>>>>>>> across the
>>>>>>>>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
>>>>>>>>>>>>>>>> still
>>>>>>>>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> As a result, I see a few possible solutions:
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> - the best and most natural seems to be extension of
>>>>>>>>>>>>>>>>> the model, so
>>>>>>>>>>>>>>>>> that it defines batch as not only "streaming pipeline
>>>>>>>>>>>>>>>>> executed in batch
>>>>>>>>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>>>>>>>>>>>>>>>>> characteristics as
>>>>>>>>>>>>>>>>> in streaming case, executed in batch fashion", I really
>>>>>>>>>>>>>>>>> don't think that
>>>>>>>>>>>>>>>>> there are any conflicts with the current model, or that this
>>>>>>>>>>>>>>>>> could
>>>>>>>>>>>>>>>>> affect performance, because the required sorting (as pointed by
>>>>>>>>>>>>>>>>> Aljoscha) is very probably already done during translation
>>>>>>>>>>>>>>>>> of stateful
>>>>>>>>>>>>>>>>> pardos. Also note that this definition only affects user
>>>>>>>>>>>>>>>>> defined
>>>>>>>>>>>>>>>>> stateful pardos
>>>>>>>>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
>>>>>>>>>>>>>>>> have
>>>>>>>>>>>>>>>> microbatch, or even a runner that alternates between
>>>>>>>>>>>>>>>> different modes.
>>>>>>>>>>>>>>>> The model describes what the valid outputs are given a
>>>>>>>>>>>>>>>> (sometimes
>>>>>>>>>>>>>>>> partial) set of inputs. It becomes really hard to define
>>>>>>>>>>>>>>>> things like
>>>>>>>>>>>>>>>> "as good runtime characteristics." Once you allow any
>>>>>>>>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
>>>>>>>>>>>>>>>> (and
>>>>>>>>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>>>>>>>>>>>>>>>> out-of-orderedness.
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
>>>>>>>>>>>>>>>> likely to
>>>>>>>>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>>>>>>>>>>>>>>>> falls
>>>>>>>>>>>>>>>> behind in streaming mode as well.
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> - another option would be to introduce annotation for
>>>>>>>>>>>>>>>>> DoFns (e.g.
>>>>>>>>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
>>>>>>>>>>>>>>>>> the sorting
>>>>>>>>>>>>>>>>> in batch case - but - this extension would have to ensure
>>>>>>>>>>>>>>>>> the sorting in
>>>>>>>>>>>>>>>>> streaming mode also - it would require definition of allowed
>>>>>>>>>>>>>>>>> lateness,
>>>>>>>>>>>>>>>>> and triggger (essentially similar to window)
>>>>>>>>>>>>>>>> This might be reasonable, implemented by default by buffering
>>>>>>>>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>>>>>>>>>>>>>>>> advances, but would likely lead to inefficient (though
>>>>>>>>>>>>>>>> *maybe* easier
>>>>>>>>>>>>>>>> to reason about) code. Not sure about the semantics of
>>>>>>>>>>>>>>>> triggering
>>>>>>>>>>>>>>>> here, especially data-driven triggers. Would it be roughly
>>>>>>>>>>>>>>>> equivalent
>>>>>>>>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
>>>>>>>>>>>>>>>> value in
>>>>>>>>>>>>>>>> values])?
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>> Or is the underlying desire just to be able to hint to the
>>>>>>>>>>>>>>>> runner that
>>>>>>>>>>>>>>>> the code may perform better (e.g. require less resources) as
>>>>>>>>>>>>>>>> skew is
>>>>>>>>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> - last option would be to introduce these "higher order
>>>>>>>>>>>>>>>>> guarantees" in
>>>>>>>>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
>>>>>>>>>>>>>>>>> worst
>>>>>>>>>>>>>>>>> option to me
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> I see the first two options quite equally good, although the
>>>>>>>>>>>>>>>>> letter one
>>>>>>>>>>>>>>>>> is probably more time consuming to implement. But it would
>>>>>>>>>>>>>>>>> bring
>>>>>>>>>>>>>>>>> additional feature to streaming case as well.
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> Thanks for any thoughts.
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> Jan
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>>>>>>>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>>>>>>>>>>>>>>>>>> <je...@seznam.cz> wrote:
>>>>>>>>>>>>>>>>>>> Hi Reuven,
>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
>>>>>>>>>>>>>>>>>>>> runners.
>>>>>>>>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
>>>>>>>>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
>>>>>>>>>>>>>>>>>>> of elements. That basically (practically) can work only
>>>>>>>>>>>>>>>>>>> for cases, where the order of input elements doesn't
>>>>>>>>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
>>>>>>>>>>>>>>>>>>> time you have a state machine involved, then the ordering
>>>>>>>>>>>>>>>>>>> of elements would matter.
>>>>>>>>>>>>>>>>>> No guarantees on order are provided in *either* streaming
>>>>>>>>>>>>>>>>>> or batch
>>>>>>>>>>>>>>>>>> mode by the model. However, it is the case that in order to
>>>>>>>>>>>>>>>>>> make
>>>>>>>>>>>>>>>>>> forward progress most streaming runners attempt to limit
>>>>>>>>>>>>>>>>>> the amount of
>>>>>>>>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
>>>>>>>>>>>>>>>>>> processing
>>>>>>>>>>>>>>>>>> time) to make forward progress, which in turn could help
>>>>>>>>>>>>>>>>>> cap the
>>>>>>>>>>>>>>>>>> amount of state that must be held concurrently, whereas a
>>>>>>>>>>>>>>>>>> batch runner
>>>>>>>>>>>>>>>>>> may not allow any state to be safely discarded until the whole
>>>>>>>>>>>>>>>>>> timeline from infinite past to infinite future has been
>>>>>>>>>>>>>>>>>> observed.
>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
>>>>>>>>>>>>>>>>>> batch" in batch mode.
>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>>>>>>>>>>>>>>>>>> <mx...@apache.org> wrote:
>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>> batch semantics and streaming semantics differs only
>>>>>>>>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
>>>>>>>>>>>>>>>>>>>> batch and cannot on stream
>>>>>>>>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
>>>>>>>>>>>>>>>>>>> trigger. You
>>>>>>>>>>>>>>>>>>> could define additional triggers that do early firings.
>>>>>>>>>>>>>>>>>>> And you could
>>>>>>>>>>>>>>>>>>> even trigger the global window by advancing the watermark
>>>>>>>>>>>>>>>>>>> to +inf.
>>>>>>>>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
>>>>>>>>>>>>>>>>>> default
>>>>>>>>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
>>>>>>>>>>>>>>>>>> is more
>>>>>>>>>>>>>>>>>> likely to be user error than an actual desire to have no
>>>>>>>>>>>>>>>>>> output until
>>>>>>>>>>>>>>>>>> drain. But it's semantically valid in the model.
Re: Definition of Unified model
Posted by Jan Lukavský <je...@seznam.cz>.
I agree, that this might be necessary. I would just think, that if we
want unified semantics, then the decision which element preceded which,
must be
a) made in realtime
b) persisted into data
That is to say - when I read data from Kafka, I can read the offset and
forward it into the pipeline for order resolution later (use custom UDF
to sort not by timestamp, but by partition offset). On the other hand
when I read from batch source, I must first have stored this "observer
dependent resolved order" into the data, otherwise, there is no way to
reconstruct it later. If it is stored in the data, then user supplied
UDF should suffice, or I am completely missing something. :)
Jan
On 5/23/19 7:17 PM, Reuven Lax wrote:
> Not really. I'm suggesting that some variant of FIFO ordering is
> necessary, which requires either runners natively support FIFO
> ordering or transforms adding some extra sequence number to each
> record to sort by.
>
> I still think your proposal is very useful by the way. I'm merely
> pointing out that to solve the state-machine problem we probably need
> something more.
>
> Reuven
>
> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
>
> Hi,
> yes. It seems that ordering by user supplied UDF makes sense and I
> will update the design proposal accordingly.
> Would that solve the issues you mention?
> Jan
> ---------- Původní e-mail ----------
> Od: Reuven Lax <relax@google.com <ma...@google.com>>
> Komu: dev <dev@beam.apache.org <ma...@beam.apache.org>>
> Datum: 23. 5. 2019 18:44:38
> Předmět: Re: Definition of Unified model
>
> I'm simply saying that timestamp ordering is insufficient for
> state machines. I wasn't proposing Kafka as a solution - that
> was simply an example of how people solve this problem in
> other scenarios.
>
> BTW another example of ordering: Imagine today that you have a
> triggered Sum aggregation writing out to a key-value sink. In
> theory we provide no ordering, so the sink might write the
> triggered sums in the wrong order, ending up with an incorrect
> value in the sink. In this case you probably want values
> ordered by trigger pane index.
>
> Reuven
>
> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
>
> Hi Reuven,
> I share the view point of Robert. I think the isuue you
> refer to is not in reality related to timestamps, but to
> the fact, that ordering of events in time is observer
> dependent (either caused by relativity, or time skew,
> essentially this has the same consequences). And the
> resolution in fact isn't Kafka, but generally an
> authoritative observer, that tells you "I saw the events
> in this order". And you either have one (and have the
> outcome of his observation persisted in the data - e.g. as
> offset in Kafka partition), then you should be able to use
> it (maybe that suggests afterall that sorting by some user
> supplied UDF might make sense), or do not have it, and
> then any interpretation of the data seems to be equally
> valid. Although determinism is fine, of course.
> Jan
> ---------- Původní e-mail ----------
> Od: Reuven Lax <relax@google.com <ma...@google.com>>
> Komu: dev <dev@beam.apache.org <ma...@beam.apache.org>>
> Datum: 23. 5. 2019 17:39:12
> Předmět: Re: Definition of Unified model
>
> So an example would be elements of type
> "startUserSession" and "endUserSession" (website
> sessions, not Beam sessions). Logically you may need
> to process them in the correct order if you have any
> sort of state-machine logic. However timestamp
> ordering is never guaranteed to match the logical
> ordering. Not only might you have several elements
> with the same timestamp, but in reality time skew
> across backend servers can cause the events to have
> timestamps in reverse order of the actual causality
> order.
>
> People do solve this problem today though. Publish the
> events to Kafka, making sure that events for the same
> user end up in the same Kafka partition. This ensures
> that the events appear in the Kafka partitions in
> causality order, even if the timestamp order doesn't
> match. The your Kafka subscriber simply process the
> elements in each partition in order.
>
> I think the ability to impose FIFO causality ordering
> is what's needed for any state-machine work. Timestamp
> ordering has advantages (though often I think the
> advantage is in state), but does not solve this problem.
>
> Reuven
>
> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw
> <robertwb@google.com <ma...@google.com>> wrote:
>
> Good point.
>
> The "implementation-specific" way I would do this is
> window-by-instant, followed by a DoFn that gets
> all the elements with
> the same timestamp and sorts/acts accordingly, but
> this counts on the
> runner producing windows in timestamp order
> (likely?) and also the
> subsequent DoFn getting them in this order (also
> likely, due to
> fusion).
>
> One could make the argument that, though it does
> not provide
> deterministic behavior, getting elements of the
> same timestamp in
> different orders should produce equally valid
> interpretations of the
> data. (After all, due to relatively, timestamps
> are not technically
> well ordered across space.) I can see how
> data-dependent tiebreakers
> could be useful, or promises of preservation of
> order between
> operations.
>
> - Robert
>
> On Thu, May 23, 2019 at 4:18 PM Reuven Lax
> <relax@google.com <ma...@google.com>> wrote:
> >
> > So Jan's example of state machines is quite a
> valid use case for ordering. However in my
> experience, timestamp ordering is insufficient for
> state machines. Elements that cause state
> transitions might come in with the exact same
> timestamp, yet still have a necessary ordering.
> Especially given Beam's decision to have
> milliseconds timestamps this is possible, but even
> at microsecond or nanosecond precision this can
> happen at scale. To handle state machines you
> usually need some sort of FIFO ordering along with
> an ordered sources, such as Kafka, not timestamp
> ordering.
> >
> > Reuven
> >
> > On Thu, May 23, 2019 at 12:32 AM Jan Lukavský
> <je.ik@seznam.cz <ma...@seznam.cz>> wrote:
> >>
> >> Hi all,
> >>
> >> thanks everyone for this discussion. I think I
> have gathered enough
> >> feedback to be able to put down a proposition
> for changes, which I will
> >> do and send to this list for further
> discussion. There are still doubts
> >> remaining the non-determinism and it's relation
> to outputs stability vs.
> >> latency. But I will try to clarify all this in
> the design document.
> >>
> >> Thanks,
> >>
> >> Jan
> >>
> >> On 5/22/19 3:49 PM, Maximilian Michels wrote:
> >> >> Someone from Flink might correct me if I'm
> wrong, but that's my
> >> >> current understanding.
> >> >
> >> > In essence your description of how
> exactly-once works in Flink is
> >> > correct. The general assumption in Flink is
> that pipelines must be
> >> > deterministic and thus produce idempotent
> writes in the case of
> >> > failures. However, that doesn't mean Beam
> sinks can't guarantee a bit
> >> > more with what Flink has to offer.
> >> >
> >> > Luke already mentioned the design discussions
> for @RequiresStableInput
> >> > which ensures idempotent writes for
> non-deterministic pipelines. This
> >> > is not part of the model but an optional Beam
> feature.
> >> >
> >> > We recently implemented support for
> @RequiresStableInput in the Flink
> >> > Runner. Reuven mentioned the Flink checkpoint
> confirmation, which
> >> > allows us to buffer (and checkpoint)
> processed data and only emit it
> >> > once a Flink checkpoint has completed.
> >> >
> >> > Cheers,
> >> > Max
> >> >
> >> > On 21.05.19 16:49, Jan Lukavský wrote:
> >> >> Hi,
> >> >>
> >> >> > Actually, I think it is a larger (open)
> question whether exactly
> >> >> once is guaranteed by the model or whether
> runners are allowed to
> >> >> relax that. I would think, however, that
> sources correctly
> >> >> implemented should be idempotent when run
> atop an exactly once
> >> >> infrastructure such as Flink of Dataflow.
> >> >>
> >> >> I would assume, that the model basically
> inherits guarantees of
> >> >> underlying infrastructure. Because Flink
> does not work as you
> >> >> described (atomic commit of inputs, state
> and outputs), but rather a
> >> >> checkpoint mark is flowing through the DAG
> much like watermark and on
> >> >> failures operators are restored and data
> reprocessed, it (IMHO)
> >> >> implies, that you have exactly once
> everywhere in the DAG *but*
> >> >> sinks. That is because sinks cannot be
> restored to previous state,
> >> >> instead sinks are supposed to be idempotent
> in order for the exactly
> >> >> once to really work (or at least be able to
> commit outputs on
> >> >> checkpoint in sink). That implies that if
> you don't have sink that is
> >> >> able to commit outputs atomically on
> checkpoint, the pipeline
> >> >> execution should be deterministic upon
> retries, otherwise shadow
> >> >> writes from failed paths of the pipeline
> might appear.
> >> >>
> >> >> Someone from Flink might correct me if I'm
> wrong, but that's my
> >> >> current understanding.
> >> >>
> >> >> > Sounds like we should make this clearer.
> >> >>
> >> >> I meant that you are right that we must not
> in any thoughts we are
> >> >> having forget that streams are by definition
> out-of-order. That is
> >> >> property that we cannot change. But - that
> doesn't limit us from
> >> >> creating operator that presents the data to
> UDF as if the stream was
> >> >> ideally sorted. It can do that by
> introducing latency, of course.
> >> >>
> >> >> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
> >> >>> Reza: One could provide something like this
> as a utility class, but
> >> >>> one downside is that it is not scale
> invariant. It requires a tuning
> >> >>> parameter that, if to small, won't mitigate
> the problem, but if to
> >> >>> big, greatly increases latency. (Possibly
> one could define a dynamic
> >> >>> session-like window to solve this
> though...) It also might be harder
> >> >>> for runners that *can* cheaply present
> stuff in timestamp order to
> >> >>> optimize. (That and, in practice, our
> annotation-style process methods
> >> >>> don't lend themselves to easy composition.)
> I think it could work in
> >> >>> specific cases though.
> >> >>>
> >> >>> More inline below.
> >> >>>
> >> >>> On Tue, May 21, 2019 at 11:38 AM Jan
> Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
> >> >>>> Hi Robert,
> >> >>>>
> >> >>>> > Beam has an exactly-once model. If the
> data was consumed, state
> >> >>>> mutated, and outputs written downstream
> (these three are committed
> >> >>>> together atomically) it will not be
> replayed. That does not, of
> >> >>>> course,
> >> >>>> solve the non-determanism due to ordering
> (including the fact that two
> >> >>>> operations reading the same PCollection
> may view different ordering).
> >> >>>>
> >> >>>> I think what you describe is a property of
> a runner, not of the model,
> >> >>>> right? I think if I run my pipeline on
> Flink I will not get this
> >> >>>> atomicity, because although Flink uses
> also exactly-once model if
> >> >>>> might
> >> >>>> write outputs multiple times.
> >> >>> Actually, I think it is a larger (open)
> question whether exactly once
> >> >>> is guaranteed by the model or whether
> runners are allowed to relax
> >> >>> that. I would think, however, that sources
> correctly implemented
> >> >>> should be idempotent when run atop an
> exactly once infrastructure such
> >> >>> as Flink of Dataflow.
> >> >>>
> >> >>>> > 1) Is it correct for a (Stateful)DoFn
> to assume elements are
> >> >>>> received
> >> >>>> in a specific order? In the current model,
> it is not. Being able to
> >> >>>> read, handle, and produced out-of-order
> data, including late data,
> >> >>>> is a
> >> >>>> pretty fundamental property of distributed
> systems.
> >> >>>>
> >> >>>> Yes, absolutely. The argument here is not
> that Stateful ParDo should
> >> >>>> presume to receive elements in any order,
> but to _present_ it as
> >> >>>> such to
> >> >>>> the user @ProcessElement function.
> >> >>> Sounds like we should make this clearer.
> >> >>>
> >> >>>> > 2) Given that some operations are
> easier (or possibly only
> >> >>>> possible)
> >> >>>> to write when operating on ordered data,
> and that different runners
> >> >>>> may
> >> >>>> have (significantly) cheaper ways to
> provide this ordering than can be
> >> >>>> done by the user themselves, should we
> elevate this to a property of
> >> >>>> (Stateful?)DoFns that the runner can
> provide? I think a compelling
> >> >>>> argument can be made here that we should.
> >> >>>>
> >> >>>> +1
> >> >>>>
> >> >>>> Jan
> >> >>>>
> >> >>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
> >> >>>>> On Mon, May 20, 2019 at 5:24 PM Jan
> Lukavský <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
> >> >>>>>> > I don't see batch vs. streaming as
> part of the model. One
> >> >>>>>> can have
> >> >>>>>> microbatch, or even a runner that
> alternates between different
> >> >>>>>> modes.
> >> >>>>>>
> >> >>>>>> Although I understand motivation of this
> statement, this project
> >> >>>>>> name is
> >> >>>>>> "Apache Beam: An advanced unified
> programming model". What does the
> >> >>>>>> model unify, if "streaming vs. batch" is
> not part of the model?
> >> >>>>> What I mean is that streaming vs. batch
> is no longer part of the
> >> >>>>> model
> >> >>>>> (or ideally API), but pushed down to be a
> concern of the runner
> >> >>>>> (executor) of the pipeline.
> >> >>>>>
> >> >>>>>
> >> >>>>> On Tue, May 21, 2019 at 10:32 AM Jan
> Lukavský <je.ik@seznam.cz <ma...@seznam.cz>>
> >> >>>>> wrote:
> >> >>>>>> Hi Kenn,
> >> >>>>>>
> >> >>>>>> OK, so if we introduce annotation, we
> can have stateful ParDo
> >> >>>>>> with sorting, that would perfectly
> resolve my issues. I still
> >> >>>>>> have some doubts, though. Let me
> explain. The current behavior of
> >> >>>>>> stateful ParDo has the following properties:
> >> >>>>>>
> >> >>>>>> a) might fail in batch, although runs
> fine in streaming (that
> >> >>>>>> is due to the buffering, and unbounded
> lateness in batch, which
> >> >>>>>> was discussed back and forth in this thread)
> >> >>>>>>
> >> >>>>>> b) might be non deterministic (this
> is because the elements
> >> >>>>>> arrive at somewhat random order, and
> even if you do the operation
> >> >>>>>> "assign unique ID to elements" this
> might produce different
> >> >>>>>> results when run multiple times)
> >> >>>>> PCollections are *explicitly* unordered.
> Any operations that
> >> >>>>> assume or
> >> >>>>> depend on a specific ordering for
> correctness (or determinism) must
> >> >>>>> provide that ordering themselves (i.e.
> tolerate "arbitrary shuffling
> >> >>>>> of inputs"). As you point out, that may
> be very expensive if you have
> >> >>>>> very hot keys with very large (unbounded)
> timestamp skew.
> >> >>>>>
> >> >>>>> StatefulDoFns are low-level operations
> that should be used with care;
> >> >>>>> the simpler windowing model gives
> determinism in the face of
> >> >>>>> unordered
> >> >>>>> data (though late data and
> non-end-of-window triggering introduces
> >> >>>>> some of the non-determanism back in).
> >> >>>>>
> >> >>>>>> What worries me most is the property b),
> because it seems to me
> >> >>>>>> to have serious consequences - not only
> that if you run twice
> >> >>>>>> batch pipeline you would get different
> results, but even on
> >> >>>>>> streaming, when pipeline fails and gets
> restarted from
> >> >>>>>> checkpoint, produced output might differ
> from the previous run
> >> >>>>>> and data from the first run might have
> already been persisted
> >> >>>>>> into sink. That would create somewhat
> messy outputs.
> >> >>>>> Beam has an exactly-once model. If the
> data was consumed, state
> >> >>>>> mutated, and outputs written downstream
> (these three are committed
> >> >>>>> together atomically) it will not be
> replayed. That does not, of
> >> >>>>> course, solve the non-determanism due to
> ordering (including the fact
> >> >>>>> that two operations reading the same
> PCollection may view different
> >> >>>>> ordering).
> >> >>>>>
> >> >>>>>> These two properties makes me think that
> the current
> >> >>>>>> implementation is more of a _special
> case_ than the general one.
> >> >>>>>> The general one would be that your state
> doesn't have the
> >> >>>>>> properties to be able to tolerate
> buffering problems and/or
> >> >>>>>> non-determinism. Which is the case where
> you need sorting in both
> >> >>>>>> streaming and batch to be part of the model.
> >> >>>>>>
> >> >>>>>> Let me point out one more analogy - that
> is merging vs.
> >> >>>>>> non-merging windows. The general case
> (merging windows) implies
> >> >>>>>> sorting by timestamp in both batch case
> (explicit) and streaming
> >> >>>>>> (buffering). The special case
> (non-merging windows) doesn't rely
> >> >>>>>> on any timestamp ordering, so the
> sorting and buffering can be
> >> >>>>>> dropped. The underlying root cause of
> this is the same for both
> >> >>>>>> stateful ParDo and windowing
> (essentially, assigning window
> >> >>>>>> labels is a stateful operation when
> windowing function is merging).
> >> >>>>>>
> >> >>>>>> The reason for the current behavior of
> stateful ParDo seems to be
> >> >>>>>> performance, but is it right to abandon
> correctness in favor of
> >> >>>>>> performance? Wouldn't it be more
> consistent to have the default
> >> >>>>>> behavior prefer correctness and when you
> have the specific
> >> >>>>>> conditions of state function having
> special properties, then you
> >> >>>>>> can annotate your DoFn (with something like
> >> >>>>>> @TimeOrderingAgnostic), which would
> yield a better performance in
> >> >>>>>> that case?
> >> >>>>> There are two separable questions here.
> >> >>>>>
> >> >>>>> 1) Is it correct for a (Stateful)DoFn to
> assume elements are received
> >> >>>>> in a specific order? In the current
> model, it is not. Being able to
> >> >>>>> read, handle, and produced out-of-order
> data, including late data, is
> >> >>>>> a pretty fundamental property of
> distributed systems.
> >> >>>>>
> >> >>>>> 2) Given that some operations are easier
> (or possibly only possible)
> >> >>>>> to write when operating on ordered data,
> and that different runners
> >> >>>>> may have (significantly) cheaper ways to
> provide this ordering than
> >> >>>>> can be done by the user themselves,
> should we elevate this to a
> >> >>>>> property of (Stateful?)DoFns that the
> runner can provide? I think a
> >> >>>>> compelling argument can be made here that
> we should.
> >> >>>>>
> >> >>>>> - Robert
> >> >>>>>
> >> >>>>>
> >> >>>>>
> >> >>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
> >> >>>>>>
> >> >>>>>> Thanks for the nice small example of a
> calculation that depends
> >> >>>>>> on order. You are right that many state
> machines have this
> >> >>>>>> property. I agree w/ you and Luke that
> it is convenient for batch
> >> >>>>>> processing to sort by event timestamp
> before running a stateful
> >> >>>>>> ParDo. In streaming you could also
> implement "sort by event
> >> >>>>>> timestamp" by buffering until you know
> all earlier data will be
> >> >>>>>> dropped - a slack buffer up to allowed
> lateness.
> >> >>>>>>
> >> >>>>>> I do not think that it is OK to sort in
> batch and not in
> >> >>>>>> streaming. Many state machines diverge
> very rapidly when things
> >> >>>>>> are out of order. So each runner if they
> see the
> >> >>>>>> "@OrderByTimestamp" annotation (or
> whatever) needs to deliver
> >> >>>>>> sorted data (by some mix of buffering
> and dropping), or to reject
> >> >>>>>> the pipeline as unsupported.
> >> >>>>>>
> >> >>>>>> And also want to say that this is not
> the default case - many
> >> >>>>>> uses of state & timers in ParDo yield
> different results at the
> >> >>>>>> element level, but the results are
> equivalent at in the big
> >> >>>>>> picture. Such as the example of "assign
> a unique sequence number
> >> >>>>>> to each element" or "group into batches"
> it doesn't matter
> >> >>>>>> exactly what the result is, only that it
> meets the spec. And
> >> >>>>>> other cases like user funnels are
> monotonic enough that you also
> >> >>>>>> don't actually need sorting.
> >> >>>>>>
> >> >>>>>> Kenn
> >> >>>>>>
> >> >>>>>> On Mon, May 20, 2019 at 2:59 PM Jan
> Lukavský <je.ik@seznam.cz <ma...@seznam.cz>>
> >> >>>>>> wrote:
> >> >>>>>>> Yes, the problem will arise probably
> mostly when you have not
> >> >>>>>>> well distributed keys (or too few
> keys). I'm really not sure if
> >> >>>>>>> a pure GBK with a trigger can solve
> this - it might help to have
> >> >>>>>>> data driven trigger. There would still
> be some doubts, though.
> >> >>>>>>> The main question is still here -
> people say, that sorting by
> >> >>>>>>> timestamp before stateful ParDo would
> be prohibitively slow, but
> >> >>>>>>> I don't really see why - the sorting is
> very probably already
> >> >>>>>>> there. And if not (hash grouping
> instead of sorted grouping),
> >> >>>>>>> then the sorting would affect only user
> defined StatefulParDos.
> >> >>>>>>>
> >> >>>>>>> This would suggest that the best way
> out of this would be really
> >> >>>>>>> to add annotation, so that the author
> of the pipeline can decide.
> >> >>>>>>>
> >> >>>>>>> If that would be acceptable I think I
> can try to prepare some
> >> >>>>>>> basic functionality, but I'm not sure,
> if I would be able to
> >> >>>>>>> cover all runners / sdks.
> >> >>>>>>>
> >> >>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
> >> >>>>>>>
> >> >>>>>>> It is read all per key and window and
> not just read all (this
> >> >>>>>>> still won't scale with hot keys in the
> global window). The GBK
> >> >>>>>>> preceding the StatefulParDo will
> guarantee that you are
> >> >>>>>>> processing all the values for a
> specific key and window at any
> >> >>>>>>> given time. Is there a specific
> window/trigger that is missing
> >> >>>>>>> that you feel would remove the need for
> you to use StatefulParDo?
> >> >>>>>>>
> >> >>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan
> Lukavský <je.ik@seznam.cz <ma...@seznam.cz>>
> >> >>>>>>> wrote:
> >> >>>>>>>> Hi Lukasz,
> >> >>>>>>>>
> >> >>>>>>>>> Today, if you must have a strict
> order, you must guarantee
> >> >>>>>>>>> that your StatefulParDo implements
> the necessary "buffering &
> >> >>>>>>>>> sorting" into state.
> >> >>>>>>>> Yes, no problem with that. But this
> whole discussion started,
> >> >>>>>>>> because *this doesn't work on batch*.
> You simply cannot first
> >> >>>>>>>> read everything from distributed
> storage and then buffer it all
> >> >>>>>>>> into memory, just to read it again,
> but sorted. That will not
> >> >>>>>>>> work. And even if it would, it would
> be a terrible waste of
> >> >>>>>>>> resources.
> >> >>>>>>>>
> >> >>>>>>>> Jan
> >> >>>>>>>>
> >> >>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
> >> >>>>>>>>
> >> >>>>>>>>
> >> >>>>>>>>
> >> >>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan
> Lukavský <je.ik@seznam.cz <ma...@seznam.cz>>
> >> >>>>>>>> wrote:
> >> >>>>>>>>> This discussion brings many really
> interesting questions for
> >> >>>>>>>>> me. :-)
> >> >>>>>>>>>
> >> >>>>>>>>> > I don't see batch vs. streaming
> as part of the model. One
> >> >>>>>>>>> can have
> >> >>>>>>>>> microbatch, or even a runner that
> alternates between different
> >> >>>>>>>>> modes.
> >> >>>>>>>>>
> >> >>>>>>>>> Although I understand motivation of
> this statement, this
> >> >>>>>>>>> project name is
> >> >>>>>>>>> "Apache Beam: An advanced unified
> programming model". What
> >> >>>>>>>>> does the
> >> >>>>>>>>> model unify, if "streaming vs. batch"
> is not part of the model?
> >> >>>>>>>>>
> >> >>>>>>>>> Using microbatching, chaining of
> batch jobs, or pure streaming
> >> >>>>>>>>> are
> >> >>>>>>>>> exactly the "runtime
> conditions/characteristics" I refer to.
> >> >>>>>>>>> All these
> >> >>>>>>>>> define several runtime parameters,
> which in turn define how
> >> >>>>>>>>> well/badly
> >> >>>>>>>>> will the pipeline perform and how
> many resources might be
> >> >>>>>>>>> needed. From
> >> >>>>>>>>> my point of view, pure streaming
> should be the most resource
> >> >>>>>>>>> demanding
> >> >>>>>>>>> (if not, why bother with batch? why
> not run everything in
> >> >>>>>>>>> streaming
> >> >>>>>>>>> only? what will there remain to
> "unify"?).
> >> >>>>>>>>>
> >> >>>>>>>>> > Fortunately, for batch, only the
> state for a single key
> >> >>>>>>>>> needs to be
> >> >>>>>>>>> preserved at a time, rather than the
> state for all keys across
> >> >>>>>>>>> the range
> >> >>>>>>>>> of skew. Of course if you have few or
> hot keys, one can still
> >> >>>>>>>>> have
> >> >>>>>>>>> issues (and this is not specific to
> StatefulDoFns).
> >> >>>>>>>>>
> >> >>>>>>>>> Yes, but here is still the
> presumption that my stateful DoFn can
> >> >>>>>>>>> tolerate arbitrary shuffling of
> inputs. Let me explain the use
> >> >>>>>>>>> case in
> >> >>>>>>>>> more detail.
> >> >>>>>>>>>
> >> >>>>>>>>> Suppose you have input stream
> consisting of 1s and 0s (and
> >> >>>>>>>>> some key for
> >> >>>>>>>>> each element, which is irrelevant for
> the demonstration). Your
> >> >>>>>>>>> task is
> >> >>>>>>>>> to calculate in running global window
> the actual number of
> >> >>>>>>>>> changes
> >> >>>>>>>>> between state 0 and state 1 and vice
> versa. When the state
> >> >>>>>>>>> doesn't
> >> >>>>>>>>> change, you don't calculate anything.
> If input (for given key)
> >> >>>>>>>>> would be
> >> >>>>>>>>> (tN denotes timestamp N):
> >> >>>>>>>>>
> >> >>>>>>>>> t1: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t2: 0
> >> >>>>>>>>>
> >> >>>>>>>>> t3: 0
> >> >>>>>>>>>
> >> >>>>>>>>> t4: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t5: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t6: 0
> >> >>>>>>>>>
> >> >>>>>>>>> then the output should yield
> (supposing that default state is
> >> >>>>>>>>> zero):
> >> >>>>>>>>>
> >> >>>>>>>>> t1: (one: 1, zero: 0)
> >> >>>>>>>>>
> >> >>>>>>>>> t2: (one: 1, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t3: (one: 1, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t4: (one: 2, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t5: (one: 2, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t6: (one: 2, zero: 2)
> >> >>>>>>>>>
> >> >>>>>>>>> How would you implement this in
> current Beam semantics?
> >> >>>>>>>> I think your saying here that I know
> that my input is ordered
> >> >>>>>>>> in a specific way and since I assume
> the order when writing my
> >> >>>>>>>> pipeline I can perform this
> optimization. But there is nothing
> >> >>>>>>>> preventing a runner from noticing that
> your processing in the
> >> >>>>>>>> global window with a specific type of
> trigger and re-ordering
> >> >>>>>>>> your inputs/processing to get better
> performance (since you
> >> >>>>>>>> can't use an AfterWatermark trigger
> for your pipeline in
> >> >>>>>>>> streaming for the GlobalWindow).
> >> >>>>>>>>
> >> >>>>>>>> Today, if you must have a strict
> order, you must guarantee that
> >> >>>>>>>> your StatefulParDo implements the
> necessary "buffering &
> >> >>>>>>>> sorting" into state. I can see why you
> would want an annotation
> >> >>>>>>>> that says I must have timestamp
> ordered elements, since it
> >> >>>>>>>> makes writing certain StatefulParDos
> much easier. StatefulParDo
> >> >>>>>>>> is a low-level function, it really is
> the "here you go and do
> >> >>>>>>>> whatever you need to but here be
> dragons" function while
> >> >>>>>>>> windowing and triggering is meant to
> keep many people from
> >> >>>>>>>> writing StatefulParDo in the first place.
> >> >>>>>>>>
> >> >>>>>>>>> > Pipelines that fail in the
> "worst case" batch scenario
> >> >>>>>>>>> are likely to
> >> >>>>>>>>> degrade poorly (possibly
> catastrophically) when the watermark
> >> >>>>>>>>> falls
> >> >>>>>>>>> behind in streaming mode as well.
> >> >>>>>>>>>
> >> >>>>>>>>> But the worst case is defined by
> input of size (available
> >> >>>>>>>>> resources +
> >> >>>>>>>>> single byte) -> pipeline fail.
> Although it could have
> >> >>>>>>>>> finished, given
> >> >>>>>>>>> the right conditions.
> >> >>>>>>>>>
> >> >>>>>>>>> > This might be reasonable,
> implemented by default by
> >> >>>>>>>>> buffering
> >> >>>>>>>>> everything and releasing elements as
> the watermark (+lateness)
> >> >>>>>>>>> advances,
> >> >>>>>>>>> but would likely lead to inefficient
> (though *maybe* easier to
> >> >>>>>>>>> reason
> >> >>>>>>>>> about) code.
> >> >>>>>>>>>
> >> >>>>>>>>> Sure, the pipeline will be less
> efficient, because it would
> >> >>>>>>>>> have to
> >> >>>>>>>>> buffer and sort the inputs. But at
> least it will produce
> >> >>>>>>>>> correct results
> >> >>>>>>>>> in cases where updates to state are
> order-sensitive.
> >> >>>>>>>>>
> >> >>>>>>>>> > Would it be roughly equivalent
> to GBK + FlatMap(lambda
> >> >>>>>>>>> (key, values):
> >> >>>>>>>>> [(key, value) for value in values])?
> >> >>>>>>>>>
> >> >>>>>>>>> I'd say roughly yes, but difference
> would be in the trigger.
> >> >>>>>>>>> The trigger
> >> >>>>>>>>> should ideally fire as soon as
> watermark (+lateness) crosses
> >> >>>>>>>>> element
> >> >>>>>>>>> with lowest timestamp in the buffer.
> Although this could be
> >> >>>>>>>>> somehow
> >> >>>>>>>>> emulated by fixed trigger each X millis.
> >> >>>>>>>>>
> >> >>>>>>>>> > Or is the underlying desire just
> to be able to hint to
> >> >>>>>>>>> the runner
> >> >>>>>>>>> that the code may perform better
> (e.g. require less resources)
> >> >>>>>>>>> as skew
> >> >>>>>>>>> is reduced (and hence to order by
> timestamp iff it's cheap)?
> >> >>>>>>>>>
> >> >>>>>>>>> No, the sorting would have to be done
> in streaming case as
> >> >>>>>>>>> well. That is
> >> >>>>>>>>> an imperative of the unified model. I
> think it is possible to
> >> >>>>>>>>> sort by
> >> >>>>>>>>> timestamp only in batch case (and do
> it for *all* batch
> >> >>>>>>>>> stateful pardos
> >> >>>>>>>>> without annotation), or introduce
> annotation, but then make
> >> >>>>>>>>> the same
> >> >>>>>>>>> guarantees for streaming case as well.
> >> >>>>>>>>>
> >> >>>>>>>>> Jan
> >> >>>>>>>>>
> >> >>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw
> wrote:
> >> >>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan
> Lukavský
> >> >>>>>>>>>> <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
> >> >>>>>>>>>>> Hi Robert,
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> yes, I think you rephrased my point
> - although no *explicit*
> >> >>>>>>>>>>> guarantees
> >> >>>>>>>>>>> of ordering are given in either
> mode, there is *implicit*
> >> >>>>>>>>>>> ordering in
> >> >>>>>>>>>>> streaming case that is due to
> nature of the processing - the
> >> >>>>>>>>>>> difference
> >> >>>>>>>>>>> between watermark and timestamp of
> elements flowing through
> >> >>>>>>>>>>> the pipeline
> >> >>>>>>>>>>> are generally low (too high
> difference leads to the
> >> >>>>>>>>>>> overbuffering
> >> >>>>>>>>>>> problem), but there is no such
> bound on batch.
> >> >>>>>>>>>> Fortunately, for batch, only the
> state for a single key needs
> >> >>>>>>>>>> to be
> >> >>>>>>>>>> preserved at a time, rather than the
> state for all keys
> >> >>>>>>>>>> across the
> >> >>>>>>>>>> range of skew. Of course if you have
> few or hot keys, one can
> >> >>>>>>>>>> still
> >> >>>>>>>>>> have issues (and this is not
> specific to StatefulDoFns).
> >> >>>>>>>>>>
> >> >>>>>>>>>>> As a result, I see a few possible
> solutions:
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> - the best and most natural
> seems to be extension of
> >> >>>>>>>>>>> the model, so
> >> >>>>>>>>>>> that it defines batch as not only
> "streaming pipeline
> >> >>>>>>>>>>> executed in batch
> >> >>>>>>>>>>> fashion", but "pipeline with at
> least as good runtime
> >> >>>>>>>>>>> characteristics as
> >> >>>>>>>>>>> in streaming case, executed in
> batch fashion", I really
> >> >>>>>>>>>>> don't think that
> >> >>>>>>>>>>> there are any conflicts with the
> current model, or that this
> >> >>>>>>>>>>> could
> >> >>>>>>>>>>> affect performance, because the
> required sorting (as pointed by
> >> >>>>>>>>>>> Aljoscha) is very probably already
> done during translation
> >> >>>>>>>>>>> of stateful
> >> >>>>>>>>>>> pardos. Also note that this
> definition only affects user
> >> >>>>>>>>>>> defined
> >> >>>>>>>>>>> stateful pardos
> >> >>>>>>>>>> I don't see batch vs. streaming as
> part of the model. One can
> >> >>>>>>>>>> have
> >> >>>>>>>>>> microbatch, or even a runner that
> alternates between
> >> >>>>>>>>>> different modes.
> >> >>>>>>>>>> The model describes what the valid
> outputs are given a
> >> >>>>>>>>>> (sometimes
> >> >>>>>>>>>> partial) set of inputs. It becomes
> really hard to define
> >> >>>>>>>>>> things like
> >> >>>>>>>>>> "as good runtime characteristics."
> Once you allow any
> >> >>>>>>>>>> out-of-orderedness, it is not very
> feasible to try and define
> >> >>>>>>>>>> (and
> >> >>>>>>>>>> more cheaply implement) a "upper
> bound" of acceptable
> >> >>>>>>>>>> out-of-orderedness.
> >> >>>>>>>>>>
> >> >>>>>>>>>> Pipelines that fail in the "worst
> case" batch scenario are
> >> >>>>>>>>>> likely to
> >> >>>>>>>>>> degrade poorly (possibly
> catastrophically) when the watermark
> >> >>>>>>>>>> falls
> >> >>>>>>>>>> behind in streaming mode as well.
> >> >>>>>>>>>>
> >> >>>>>>>>>>> - another option would be to
> introduce annotation for
> >> >>>>>>>>>>> DoFns (e.g.
> >> >>>>>>>>>>>
> @RequiresStableTimeCharacteristics), which would
> result in
> >> >>>>>>>>>>> the sorting
> >> >>>>>>>>>>> in batch case - but - this
> extension would have to ensure
> >> >>>>>>>>>>> the sorting in
> >> >>>>>>>>>>> streaming mode also - it would
> require definition of allowed
> >> >>>>>>>>>>> lateness,
> >> >>>>>>>>>>> and triggger (essentially similar
> to window)
> >> >>>>>>>>>> This might be reasonable,
> implemented by default by buffering
> >> >>>>>>>>>> everything and releasing elements as
> the watermark (+lateness)
> >> >>>>>>>>>> advances, but would likely lead to
> inefficient (though
> >> >>>>>>>>>> *maybe* easier
> >> >>>>>>>>>> to reason about) code. Not sure
> about the semantics of
> >> >>>>>>>>>> triggering
> >> >>>>>>>>>> here, especially data-driven
> triggers. Would it be roughly
> >> >>>>>>>>>> equivalent
> >> >>>>>>>>>> to GBK + FlatMap(lambda (key,
> values): [(key, value) for
> >> >>>>>>>>>> value in
> >> >>>>>>>>>> values])?
> >> >>>>>>>>>>
> >> >>>>>>>>>> Or is the underlying desire just to
> be able to hint to the
> >> >>>>>>>>>> runner that
> >> >>>>>>>>>> the code may perform better (e.g.
> require less resources) as
> >> >>>>>>>>>> skew is
> >> >>>>>>>>>> reduced (and hence to order by
> timestamp iff it's cheap)?
> >> >>>>>>>>>>
> >> >>>>>>>>>>> - last option would be to
> introduce these "higher order
> >> >>>>>>>>>>> guarantees" in
> >> >>>>>>>>>>> some extension DSL (e.g. Euphoria),
> but that seems to be the
> >> >>>>>>>>>>> worst
> >> >>>>>>>>>>> option to me
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> I see the first two options quite
> equally good, although the
> >> >>>>>>>>>>> letter one
> >> >>>>>>>>>>> is probably more time consuming to
> implement. But it would
> >> >>>>>>>>>>> bring
> >> >>>>>>>>>>> additional feature to streaming
> case as well.
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> Thanks for any thoughts.
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> Jan
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> On 5/20/19 12:41 PM, Robert
> Bradshaw wrote:
> >> >>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM
> Jan Lukavský
> >> >>>>>>>>>>>> <je.ik@seznam.cz
> <ma...@seznam.cz>> wrote:
> >> >>>>>>>>>>>>> Hi Reuven,
> >> >>>>>>>>>>>>>
> >> >>>>>>>>>>>>>> How so? AFAIK stateful DoFns
> work just fine in batch
> >> >>>>>>>>>>>>>> runners.
> >> >>>>>>>>>>>>> Stateful ParDo works in batch as
> far, as the logic inside
> >> >>>>>>>>>>>>> the state works for absolutely
> unbounded out-of-orderness
> >> >>>>>>>>>>>>> of elements. That basically
> (practically) can work only
> >> >>>>>>>>>>>>> for cases, where the order of
> input elements doesn't
> >> >>>>>>>>>>>>> matter. But, "state" can refer to
> "state machine", and any
> >> >>>>>>>>>>>>> time you have a state machine
> involved, then the ordering
> >> >>>>>>>>>>>>> of elements would matter.
> >> >>>>>>>>>>>> No guarantees on order are
> provided in *either* streaming
> >> >>>>>>>>>>>> or batch
> >> >>>>>>>>>>>> mode by the model. However, it is
> the case that in order to
> >> >>>>>>>>>>>> make
> >> >>>>>>>>>>>> forward progress most streaming
> runners attempt to limit
> >> >>>>>>>>>>>> the amount of
> >> >>>>>>>>>>>> out-of-orderedness of elements (in
> terms of event time vs.
> >> >>>>>>>>>>>> processing
> >> >>>>>>>>>>>> time) to make forward progress,
> which in turn could help
> >> >>>>>>>>>>>> cap the
> >> >>>>>>>>>>>> amount of state that must be held
> concurrently, whereas a
> >> >>>>>>>>>>>> batch runner
> >> >>>>>>>>>>>> may not allow any state to be
> safely discarded until the whole
> >> >>>>>>>>>>>> timeline from infinite past to
> infinite future has been
> >> >>>>>>>>>>>> observed.
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>> Also, as pointed out, state is not
> preserved "batch to
> >> >>>>>>>>>>>> batch" in batch mode.
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM
> Maximilian Michels
> >> >>>>>>>>>>>> <mxm@apache.org
> <ma...@apache.org>> wrote:
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>>>> batch semantics and
> streaming semantics differs only
> >> >>>>>>>>>>>>>> in that I can have GlobalWindow
> with default trigger on
> >> >>>>>>>>>>>>>> batch and cannot on stream
> >> >>>>>>>>>>>>> You can have a GlobalWindow in
> streaming with a default
> >> >>>>>>>>>>>>> trigger. You
> >> >>>>>>>>>>>>> could define additional triggers
> that do early firings.
> >> >>>>>>>>>>>>> And you could
> >> >>>>>>>>>>>>> even trigger the global window by
> advancing the watermark
> >> >>>>>>>>>>>>> to +inf.
> >> >>>>>>>>>>>> IIRC, as a pragmatic note, we
> prohibited global window with
> >> >>>>>>>>>>>> default
> >> >>>>>>>>>>>> trigger on unbounded PCollections
> in the SDK because this
> >> >>>>>>>>>>>> is more
> >> >>>>>>>>>>>> likely to be user error than an
> actual desire to have no
> >> >>>>>>>>>>>> output until
> >> >>>>>>>>>>>> drain. But it's semantically valid
> in the model.
>
Re: Definition of Unified model
Posted by Lukasz Cwik <lc...@google.com>.
On Tue, May 21, 2019 at 7:49 AM Jan Lukavský <je...@seznam.cz> wrote:
> Hi,
>
> > Actually, I think it is a larger (open) question whether exactly once
> is guaranteed by the model or whether runners are allowed to relax that.
> I would think, however, that sources correctly implemented should be
> idempotent when run atop an exactly once infrastructure such as Flink of
> Dataflow.
>
Exactly once semantics is not required in the Beam model. There is a
TODO[1] on the runners capability matrix to expand it stating which runner
provides at least once vs exactly once and other runtime characteristics.
> I would assume, that the model basically inherits guarantees of
> underlying infrastructure. Because Flink does not work as you described
> (atomic commit of inputs, state and outputs), but rather a checkpoint
> mark is flowing through the DAG much like watermark and on failures
> operators are restored and data reprocessed, it (IMHO) implies, that you
> have exactly once everywhere in the DAG *but* sinks. That is because
> sinks cannot be restored to previous state, instead sinks are supposed
> to be idempotent in order for the exactly once to really work (or at
> least be able to commit outputs on checkpoint in sink). That implies
> that if you don't have sink that is able to commit outputs atomically on
> checkpoint, the pipeline execution should be deterministic upon retries,
> otherwise shadow writes from failed paths of the pipeline might appear.
>
There was a discussion about transforms that require stable input to
guarantee correct semantics. This requirement usually came up when
attempting to write a sink correctly. More details on it in these two
threads [2, 3].
> Someone from Flink might correct me if I'm wrong, but that's my current
> understanding.
>
> > Sounds like we should make this clearer.
>
> I meant that you are right that we must not in any thoughts we are
> having forget that streams are by definition out-of-order. That is
> property that we cannot change. But - that doesn't limit us from
> creating operator that presents the data to UDF as if the stream was
> ideally sorted. It can do that by introducing latency, of course.
>
> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
> > Reza: One could provide something like this as a utility class, but
> > one downside is that it is not scale invariant. It requires a tuning
> > parameter that, if to small, won't mitigate the problem, but if to
> > big, greatly increases latency. (Possibly one could define a dynamic
> > session-like window to solve this though...) It also might be harder
> > for runners that *can* cheaply present stuff in timestamp order to
> > optimize. (That and, in practice, our annotation-style process methods
> > don't lend themselves to easy composition.) I think it could work in
> > specific cases though.
> >
> > More inline below.
> >
> > On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz> wrote:
> >> Hi Robert,
> >>
> >> > Beam has an exactly-once model. If the data was consumed, state
> >> mutated, and outputs written downstream (these three are committed
> >> together atomically) it will not be replayed. That does not, of course,
> >> solve the non-determanism due to ordering (including the fact that two
> >> operations reading the same PCollection may view different ordering).
> >>
> >> I think what you describe is a property of a runner, not of the model,
> >> right? I think if I run my pipeline on Flink I will not get this
> >> atomicity, because although Flink uses also exactly-once model if might
> >> write outputs multiple times.
> > Actually, I think it is a larger (open) question whether exactly once
> > is guaranteed by the model or whether runners are allowed to relax
> > that. I would think, however, that sources correctly implemented
> > should be idempotent when run atop an exactly once infrastructure such
> > as Flink of Dataflow.
> >
> >> > 1) Is it correct for a (Stateful)DoFn to assume elements are
> received
> >> in a specific order? In the current model, it is not. Being able to
> >> read, handle, and produced out-of-order data, including late data, is a
> >> pretty fundamental property of distributed systems.
> >>
> >> Yes, absolutely. The argument here is not that Stateful ParDo should
> >> presume to receive elements in any order, but to _present_ it as such to
> >> the user @ProcessElement function.
> > Sounds like we should make this clearer.
> >
> >> > 2) Given that some operations are easier (or possibly only possible)
> >> to write when operating on ordered data, and that different runners may
> >> have (significantly) cheaper ways to provide this ordering than can be
> >> done by the user themselves, should we elevate this to a property of
> >> (Stateful?)DoFns that the runner can provide? I think a compelling
> >> argument can be made here that we should.
> >>
> >> +1
> >>
> >> Jan
> >>
> >> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
> >>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz> wrote:
> >>>> > I don't see batch vs. streaming as part of the model. One can
> have
> >>>> microbatch, or even a runner that alternates between different modes.
> >>>>
> >>>> Although I understand motivation of this statement, this project name
> is
> >>>> "Apache Beam: An advanced unified programming model". What does the
> >>>> model unify, if "streaming vs. batch" is not part of the model?
> >>> What I mean is that streaming vs. batch is no longer part of the model
> >>> (or ideally API), but pushed down to be a concern of the runner
> >>> (executor) of the pipeline.
> >>>
> >>>
> >>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz> wrote:
> >>>> Hi Kenn,
> >>>>
> >>>> OK, so if we introduce annotation, we can have stateful ParDo with
> sorting, that would perfectly resolve my issues. I still have some doubts,
> though. Let me explain. The current behavior of stateful ParDo has the
> following properties:
> >>>>
> >>>> a) might fail in batch, although runs fine in streaming (that is
> due to the buffering, and unbounded lateness in batch, which was discussed
> back and forth in this thread)
> >>>>
> >>>> b) might be non deterministic (this is because the elements arrive
> at somewhat random order, and even if you do the operation "assign unique
> ID to elements" this might produce different results when run multiple
> times)
> >>> PCollections are *explicitly* unordered. Any operations that assume or
> >>> depend on a specific ordering for correctness (or determinism) must
> >>> provide that ordering themselves (i.e. tolerate "arbitrary shuffling
> >>> of inputs"). As you point out, that may be very expensive if you have
> >>> very hot keys with very large (unbounded) timestamp skew.
> >>>
> >>> StatefulDoFns are low-level operations that should be used with care;
> >>> the simpler windowing model gives determinism in the face of unordered
> >>> data (though late data and non-end-of-window triggering introduces
> >>> some of the non-determanism back in).
> >>>
> >>>> What worries me most is the property b), because it seems to me to
> have serious consequences - not only that if you run twice batch pipeline
> you would get different results, but even on streaming, when pipeline fails
> and gets restarted from checkpoint, produced output might differ from the
> previous run and data from the first run might have already been persisted
> into sink. That would create somewhat messy outputs.
> >>> Beam has an exactly-once model. If the data was consumed, state
> >>> mutated, and outputs written downstream (these three are committed
> >>> together atomically) it will not be replayed. That does not, of
> >>> course, solve the non-determanism due to ordering (including the fact
> >>> that two operations reading the same PCollection may view different
> >>> ordering).
> >>>
> >>>> These two properties makes me think that the current implementation
> is more of a _special case_ than the general one. The general one would be
> that your state doesn't have the properties to be able to tolerate
> buffering problems and/or non-determinism. Which is the case where you need
> sorting in both streaming and batch to be part of the model.
> >>>>
> >>>> Let me point out one more analogy - that is merging vs. non-merging
> windows. The general case (merging windows) implies sorting by timestamp in
> both batch case (explicit) and streaming (buffering). The special case
> (non-merging windows) doesn't rely on any timestamp ordering, so the
> sorting and buffering can be dropped. The underlying root cause of this is
> the same for both stateful ParDo and windowing (essentially, assigning
> window labels is a stateful operation when windowing function is merging).
> >>>>
> >>>> The reason for the current behavior of stateful ParDo seems to be
> performance, but is it right to abandon correctness in favor of
> performance? Wouldn't it be more consistent to have the default behavior
> prefer correctness and when you have the specific conditions of state
> function having special properties, then you can annotate your DoFn (with
> something like @TimeOrderingAgnostic), which would yield a better
> performance in that case?
> >>> There are two separable questions here.
> >>>
> >>> 1) Is it correct for a (Stateful)DoFn to assume elements are received
> >>> in a specific order? In the current model, it is not. Being able to
> >>> read, handle, and produced out-of-order data, including late data, is
> >>> a pretty fundamental property of distributed systems.
> >>>
> >>> 2) Given that some operations are easier (or possibly only possible)
> >>> to write when operating on ordered data, and that different runners
> >>> may have (significantly) cheaper ways to provide this ordering than
> >>> can be done by the user themselves, should we elevate this to a
> >>> property of (Stateful?)DoFns that the runner can provide? I think a
> >>> compelling argument can be made here that we should.
> >>>
> >>> - Robert
> >>>
> >>>
> >>>
> >>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
> >>>>
> >>>> Thanks for the nice small example of a calculation that depends on
> order. You are right that many state machines have this property. I agree
> w/ you and Luke that it is convenient for batch processing to sort by event
> timestamp before running a stateful ParDo. In streaming you could also
> implement "sort by event timestamp" by buffering until you know all earlier
> data will be dropped - a slack buffer up to allowed lateness.
> >>>>
> >>>> I do not think that it is OK to sort in batch and not in streaming.
> Many state machines diverge very rapidly when things are out of order. So
> each runner if they see the "@OrderByTimestamp" annotation (or whatever)
> needs to deliver sorted data (by some mix of buffering and dropping), or to
> reject the pipeline as unsupported.
> >>>>
> >>>> And also want to say that this is not the default case - many uses of
> state & timers in ParDo yield different results at the element level, but
> the results are equivalent at in the big picture. Such as the example of
> "assign a unique sequence number to each element" or "group into batches"
> it doesn't matter exactly what the result is, only that it meets the spec.
> And other cases like user funnels are monotonic enough that you also don't
> actually need sorting.
> >>>>
> >>>> Kenn
> >>>>
> >>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz> wrote:
> >>>>> Yes, the problem will arise probably mostly when you have not well
> distributed keys (or too few keys). I'm really not sure if a pure GBK with
> a trigger can solve this - it might help to have data driven trigger. There
> would still be some doubts, though. The main question is still here -
> people say, that sorting by timestamp before stateful ParDo would be
> prohibitively slow, but I don't really see why - the sorting is very
> probably already there. And if not (hash grouping instead of sorted
> grouping), then the sorting would affect only user defined StatefulParDos.
> >>>>>
> >>>>> This would suggest that the best way out of this would be really to
> add annotation, so that the author of the pipeline can decide.
> >>>>>
> >>>>> If that would be acceptable I think I can try to prepare some basic
> functionality, but I'm not sure, if I would be able to cover all runners /
> sdks.
> >>>>>
> >>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
> >>>>>
> >>>>> It is read all per key and window and not just read all (this still
> won't scale with hot keys in the global window). The GBK preceding the
> StatefulParDo will guarantee that you are processing all the values for a
> specific key and window at any given time. Is there a specific
> window/trigger that is missing that you feel would remove the need for you
> to use StatefulParDo?
> >>>>>
> >>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>> Hi Lukasz,
> >>>>>>
> >>>>>>> Today, if you must have a strict order, you must guarantee that
> your StatefulParDo implements the necessary "buffering & sorting" into
> state.
> >>>>>> Yes, no problem with that. But this whole discussion started,
> because *this doesn't work on batch*. You simply cannot first read
> everything from distributed storage and then buffer it all into memory,
> just to read it again, but sorted. That will not work. And even if it
> would, it would be a terrible waste of resources.
> >>>>>>
> >>>>>> Jan
> >>>>>>
> >>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>>> This discussion brings many really interesting questions for me.
> :-)
> >>>>>>>
> >>>>>>> > I don't see batch vs. streaming as part of the model. One can
> have
> >>>>>>> microbatch, or even a runner that alternates between different
> modes.
> >>>>>>>
> >>>>>>> Although I understand motivation of this statement, this project
> name is
> >>>>>>> "Apache Beam: An advanced unified programming model". What does the
> >>>>>>> model unify, if "streaming vs. batch" is not part of the model?
> >>>>>>>
> >>>>>>> Using microbatching, chaining of batch jobs, or pure streaming are
> >>>>>>> exactly the "runtime conditions/characteristics" I refer to. All
> these
> >>>>>>> define several runtime parameters, which in turn define how
> well/badly
> >>>>>>> will the pipeline perform and how many resources might be needed.
> From
> >>>>>>> my point of view, pure streaming should be the most resource
> demanding
> >>>>>>> (if not, why bother with batch? why not run everything in streaming
> >>>>>>> only? what will there remain to "unify"?).
> >>>>>>>
> >>>>>>> > Fortunately, for batch, only the state for a single key needs
> to be
> >>>>>>> preserved at a time, rather than the state for all keys across the
> range
> >>>>>>> of skew. Of course if you have few or hot keys, one can still have
> >>>>>>> issues (and this is not specific to StatefulDoFns).
> >>>>>>>
> >>>>>>> Yes, but here is still the presumption that my stateful DoFn can
> >>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
> case in
> >>>>>>> more detail.
> >>>>>>>
> >>>>>>> Suppose you have input stream consisting of 1s and 0s (and some
> key for
> >>>>>>> each element, which is irrelevant for the demonstration). Your
> task is
> >>>>>>> to calculate in running global window the actual number of changes
> >>>>>>> between state 0 and state 1 and vice versa. When the state doesn't
> >>>>>>> change, you don't calculate anything. If input (for given key)
> would be
> >>>>>>> (tN denotes timestamp N):
> >>>>>>>
> >>>>>>> t1: 1
> >>>>>>>
> >>>>>>> t2: 0
> >>>>>>>
> >>>>>>> t3: 0
> >>>>>>>
> >>>>>>> t4: 1
> >>>>>>>
> >>>>>>> t5: 1
> >>>>>>>
> >>>>>>> t6: 0
> >>>>>>>
> >>>>>>> then the output should yield (supposing that default state is
> zero):
> >>>>>>>
> >>>>>>> t1: (one: 1, zero: 0)
> >>>>>>>
> >>>>>>> t2: (one: 1, zero: 1)
> >>>>>>>
> >>>>>>> t3: (one: 1, zero: 1)
> >>>>>>>
> >>>>>>> t4: (one: 2, zero: 1)
> >>>>>>>
> >>>>>>> t5: (one: 2, zero: 1)
> >>>>>>>
> >>>>>>> t6: (one: 2, zero: 2)
> >>>>>>>
> >>>>>>> How would you implement this in current Beam semantics?
> >>>>>> I think your saying here that I know that my input is ordered in a
> specific way and since I assume the order when writing my pipeline I can
> perform this optimization. But there is nothing preventing a runner from
> noticing that your processing in the global window with a specific type of
> trigger and re-ordering your inputs/processing to get better performance
> (since you can't use an AfterWatermark trigger for your pipeline in
> streaming for the GlobalWindow).
> >>>>>>
> >>>>>> Today, if you must have a strict order, you must guarantee that
> your StatefulParDo implements the necessary "buffering & sorting" into
> state. I can see why you would want an annotation that says I must have
> timestamp ordered elements, since it makes writing certain StatefulParDos
> much easier. StatefulParDo is a low-level function, it really is the "here
> you go and do whatever you need to but here be dragons" function while
> windowing and triggering is meant to keep many people from writing
> StatefulParDo in the first place.
> >>>>>>
> >>>>>>> > Pipelines that fail in the "worst case" batch scenario are
> likely to
> >>>>>>> degrade poorly (possibly catastrophically) when the watermark falls
> >>>>>>> behind in streaming mode as well.
> >>>>>>>
> >>>>>>> But the worst case is defined by input of size (available
> resources +
> >>>>>>> single byte) -> pipeline fail. Although it could have finished,
> given
> >>>>>>> the right conditions.
> >>>>>>>
> >>>>>>> > This might be reasonable, implemented by default by buffering
> >>>>>>> everything and releasing elements as the watermark (+lateness)
> advances,
> >>>>>>> but would likely lead to inefficient (though *maybe* easier to
> reason
> >>>>>>> about) code.
> >>>>>>>
> >>>>>>> Sure, the pipeline will be less efficient, because it would have to
> >>>>>>> buffer and sort the inputs. But at least it will produce correct
> results
> >>>>>>> in cases where updates to state are order-sensitive.
> >>>>>>>
> >>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda (key,
> values):
> >>>>>>> [(key, value) for value in values])?
> >>>>>>>
> >>>>>>> I'd say roughly yes, but difference would be in the trigger. The
> trigger
> >>>>>>> should ideally fire as soon as watermark (+lateness) crosses
> element
> >>>>>>> with lowest timestamp in the buffer. Although this could be somehow
> >>>>>>> emulated by fixed trigger each X millis.
> >>>>>>>
> >>>>>>> > Or is the underlying desire just to be able to hint to the
> runner
> >>>>>>> that the code may perform better (e.g. require less resources) as
> skew
> >>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
> >>>>>>>
> >>>>>>> No, the sorting would have to be done in streaming case as well.
> That is
> >>>>>>> an imperative of the unified model. I think it is possible to sort
> by
> >>>>>>> timestamp only in batch case (and do it for *all* batch stateful
> pardos
> >>>>>>> without annotation), or introduce annotation, but then make the
> same
> >>>>>>> guarantees for streaming case as well.
> >>>>>>>
> >>>>>>> Jan
> >>>>>>>
> >>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
> >>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>>>>> Hi Robert,
> >>>>>>>>>
> >>>>>>>>> yes, I think you rephrased my point - although no *explicit*
> guarantees
> >>>>>>>>> of ordering are given in either mode, there is *implicit*
> ordering in
> >>>>>>>>> streaming case that is due to nature of the processing - the
> difference
> >>>>>>>>> between watermark and timestamp of elements flowing through the
> pipeline
> >>>>>>>>> are generally low (too high difference leads to the overbuffering
> >>>>>>>>> problem), but there is no such bound on batch.
> >>>>>>>> Fortunately, for batch, only the state for a single key needs to
> be
> >>>>>>>> preserved at a time, rather than the state for all keys across the
> >>>>>>>> range of skew. Of course if you have few or hot keys, one can
> still
> >>>>>>>> have issues (and this is not specific to StatefulDoFns).
> >>>>>>>>
> >>>>>>>>> As a result, I see a few possible solutions:
> >>>>>>>>>
> >>>>>>>>> - the best and most natural seems to be extension of the
> model, so
> >>>>>>>>> that it defines batch as not only "streaming pipeline executed
> in batch
> >>>>>>>>> fashion", but "pipeline with at least as good runtime
> characteristics as
> >>>>>>>>> in streaming case, executed in batch fashion", I really don't
> think that
> >>>>>>>>> there are any conflicts with the current model, or that this
> could
> >>>>>>>>> affect performance, because the required sorting (as pointed by
> >>>>>>>>> Aljoscha) is very probably already done during translation of
> stateful
> >>>>>>>>> pardos. Also note that this definition only affects user defined
> >>>>>>>>> stateful pardos
> >>>>>>>> I don't see batch vs. streaming as part of the model. One can have
> >>>>>>>> microbatch, or even a runner that alternates between different
> modes.
> >>>>>>>> The model describes what the valid outputs are given a (sometimes
> >>>>>>>> partial) set of inputs. It becomes really hard to define things
> like
> >>>>>>>> "as good runtime characteristics." Once you allow any
> >>>>>>>> out-of-orderedness, it is not very feasible to try and define (and
> >>>>>>>> more cheaply implement) a "upper bound" of acceptable
> >>>>>>>> out-of-orderedness.
> >>>>>>>>
> >>>>>>>> Pipelines that fail in the "worst case" batch scenario are likely
> to
> >>>>>>>> degrade poorly (possibly catastrophically) when the watermark
> falls
> >>>>>>>> behind in streaming mode as well.
> >>>>>>>>
> >>>>>>>>> - another option would be to introduce annotation for DoFns
> (e.g.
> >>>>>>>>> @RequiresStableTimeCharacteristics), which would result in the
> sorting
> >>>>>>>>> in batch case - but - this extension would have to ensure the
> sorting in
> >>>>>>>>> streaming mode also - it would require definition of allowed
> lateness,
> >>>>>>>>> and triggger (essentially similar to window)
> >>>>>>>> This might be reasonable, implemented by default by buffering
> >>>>>>>> everything and releasing elements as the watermark (+lateness)
> >>>>>>>> advances, but would likely lead to inefficient (though *maybe*
> easier
> >>>>>>>> to reason about) code. Not sure about the semantics of triggering
> >>>>>>>> here, especially data-driven triggers. Would it be roughly
> equivalent
> >>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for value in
> >>>>>>>> values])?
> >>>>>>>>
> >>>>>>>> Or is the underlying desire just to be able to hint to the runner
> that
> >>>>>>>> the code may perform better (e.g. require less resources) as skew
> is
> >>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
> >>>>>>>>
> >>>>>>>>> - last option would be to introduce these "higher order
> guarantees" in
> >>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
> worst
> >>>>>>>>> option to me
> >>>>>>>>>
> >>>>>>>>> I see the first two options quite equally good, although the
> letter one
> >>>>>>>>> is probably more time consuming to implement. But it would bring
> >>>>>>>>> additional feature to streaming case as well.
> >>>>>>>>>
> >>>>>>>>> Thanks for any thoughts.
> >>>>>>>>>
> >>>>>>>>> Jan
> >>>>>>>>>
> >>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
> >>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>>>>>>> Hi Reuven,
> >>>>>>>>>>>
> >>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch runners.
> >>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside the
> state works for absolutely unbounded out-of-orderness of elements. That
> basically (practically) can work only for cases, where the order of input
> elements doesn't matter. But, "state" can refer to "state machine", and any
> time you have a state machine involved, then the ordering of elements would
> matter.
> >>>>>>>>>> No guarantees on order are provided in *either* streaming or
> batch
> >>>>>>>>>> mode by the model. However, it is the case that in order to make
> >>>>>>>>>> forward progress most streaming runners attempt to limit the
> amount of
> >>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
> processing
> >>>>>>>>>> time) to make forward progress, which in turn could help cap the
> >>>>>>>>>> amount of state that must be held concurrently, whereas a batch
> runner
> >>>>>>>>>> may not allow any state to be safely discarded until the whole
> >>>>>>>>>> timeline from infinite past to infinite future has been
> observed.
> >>>>>>>>>>
> >>>>>>>>>> Also, as pointed out, state is not preserved "batch to batch"
> in batch mode.
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels <
> mxm@apache.org> wrote:
> >>>>>>>>>>
> >>>>>>>>>>>> batch semantics and streaming semantics differs only in
> that I can have GlobalWindow with default trigger on batch and cannot on
> stream
> >>>>>>>>>>> You can have a GlobalWindow in streaming with a default
> trigger. You
> >>>>>>>>>>> could define additional triggers that do early firings. And
> you could
> >>>>>>>>>>> even trigger the global window by advancing the watermark to
> +inf.
> >>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
> default
> >>>>>>>>>> trigger on unbounded PCollections in the SDK because this is
> more
> >>>>>>>>>> likely to be user error than an actual desire to have no output
> until
> >>>>>>>>>> drain. But it's semantically valid in the model.
>
1: https://beam.apache.org/documentation/runners/capability-matrix/
2:
https://lists.apache.org/thread.html/7487750085413f11fbeb1e707fa8fc138a07c93d6a394cf7245371fc@%3Cdev.beam.apache.org%3E
3:
https://lists.apache.org/thread.html/d0c24cb4d421e25980fee1855d1daa68d343700270ed962d642c8f99@%3Cdev.beam.apache.org%3E
Re: Definition of Unified model
Posted by Robert Bradshaw <ro...@google.com>.
Huge +1 to all Kenn said.
Jan, batch sources can have orderings too, just like Kafka. I think
it's reasonable (for both batch and streaming) that if a source has an
ordering that is an important part of the data, it should preserve
this ordering into the data itself (e.g. as sequence numbers, offsets,
etc.)
On Fri, May 24, 2019 at 10:35 PM Kenneth Knowles <ke...@apache.org> wrote:
>
> I strongly prefer explicit sequence metadata over FIFO requirements, because:
>
> - FIFO is complex to specify: for example Dataflow has "per stage key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage" is not a portable concept, nor even guaranteed to remain a Dataflow concept)
> - complex specifications are by definition poor usability (if necessary, then it is what it is)
> - overly restricts the runner, reduces parallelism, for example any non-stateful ParDo has per-element parallelism, not per "key"
> - another perspective on that: FIFO makes everyone pay rather than just the transform that requires exactly sequencing
> - previous implementation details like reshuffles become part of the model
> - I'm not even convinced the use cases involved are addressed by some careful FIFO restrictions; many sinks re-key and they would all have to become aware of how keying of a sequence of "stages" affects the end-to-end FIFO
>
> A noop becoming a non-noop is essentially the mathematical definition of moving from higher-level to lower-level abstraction.
>
> So this strikes at the core question of what level of abstraction Beam aims to represent. Lower-level means there are fewer possible implementations and it is more tied to the underlying architecture, and anything not near-exact match pays a huge penalty. Higher-level means there are more implementations possible with different tradeoffs, though they may all pay a minor penalty.
>
> I could be convinced to change my mind, but it needs some extensive design, examples, etc. I think it is probably about the most consequential design decision in the whole Beam model, around the same level as the decision to use ParDo and GBK as the primitives IMO.
>
> Kenn
>
> On Thu, May 23, 2019 at 10:17 AM Reuven Lax <re...@google.com> wrote:
>>
>> Not really. I'm suggesting that some variant of FIFO ordering is necessary, which requires either runners natively support FIFO ordering or transforms adding some extra sequence number to each record to sort by.
>>
>> I still think your proposal is very useful by the way. I'm merely pointing out that to solve the state-machine problem we probably need something more.
>>
>> Reuven
>>
>> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>
>>> Hi,
>>> yes. It seems that ordering by user supplied UDF makes sense and I will update the design proposal accordingly.
>>> Would that solve the issues you mention?
>>> Jan
>>> ---------- Původní e-mail ----------
>>> Od: Reuven Lax <re...@google.com>
>>> Komu: dev <de...@beam.apache.org>
>>> Datum: 23. 5. 2019 18:44:38
>>> Předmět: Re: Definition of Unified model
>>>
>>> I'm simply saying that timestamp ordering is insufficient for state machines. I wasn't proposing Kafka as a solution - that was simply an example of how people solve this problem in other scenarios.
>>>
>>> BTW another example of ordering: Imagine today that you have a triggered Sum aggregation writing out to a key-value sink. In theory we provide no ordering, so the sink might write the triggered sums in the wrong order, ending up with an incorrect value in the sink. In this case you probably want values ordered by trigger pane index.
>>>
>>> Reuven
>>>
>>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>
>>> Hi Reuven,
>>> I share the view point of Robert. I think the isuue you refer to is not in reality related to timestamps, but to the fact, that ordering of events in time is observer dependent (either caused by relativity, or time skew, essentially this has the same consequences). And the resolution in fact isn't Kafka, but generally an authoritative observer, that tells you "I saw the events in this order". And you either have one (and have the outcome of his observation persisted in the data - e.g. as offset in Kafka partition), then you should be able to use it (maybe that suggests afterall that sorting by some user supplied UDF might make sense), or do not have it, and then any interpretation of the data seems to be equally valid. Although determinism is fine, of course.
>>> Jan
>>> ---------- Původní e-mail ----------
>>> Od: Reuven Lax <re...@google.com>
>>> Komu: dev <de...@beam.apache.org>
>>> Datum: 23. 5. 2019 17:39:12
>>> Předmět: Re: Definition of Unified model
>>>
>>> So an example would be elements of type "startUserSession" and "endUserSession" (website sessions, not Beam sessions). Logically you may need to process them in the correct order if you have any sort of state-machine logic. However timestamp ordering is never guaranteed to match the logical ordering. Not only might you have several elements with the same timestamp, but in reality time skew across backend servers can cause the events to have timestamps in reverse order of the actual causality order.
>>>
>>> People do solve this problem today though. Publish the events to Kafka, making sure that events for the same user end up in the same Kafka partition. This ensures that the events appear in the Kafka partitions in causality order, even if the timestamp order doesn't match. The your Kafka subscriber simply process the elements in each partition in order.
>>>
>>> I think the ability to impose FIFO causality ordering is what's needed for any state-machine work. Timestamp ordering has advantages (though often I think the advantage is in state), but does not solve this problem.
>>>
>>> Reuven
>>>
>>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <ro...@google.com> wrote:
>>>
>>> Good point.
>>>
>>> The "implementation-specific" way I would do this is
>>> window-by-instant, followed by a DoFn that gets all the elements with
>>> the same timestamp and sorts/acts accordingly, but this counts on the
>>> runner producing windows in timestamp order (likely?) and also the
>>> subsequent DoFn getting them in this order (also likely, due to
>>> fusion).
>>>
>>> One could make the argument that, though it does not provide
>>> deterministic behavior, getting elements of the same timestamp in
>>> different orders should produce equally valid interpretations of the
>>> data. (After all, due to relatively, timestamps are not technically
>>> well ordered across space.) I can see how data-dependent tiebreakers
>>> could be useful, or promises of preservation of order between
>>> operations.
>>>
>>> - Robert
>>>
>>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com> wrote:
>>> >
>>> > So Jan's example of state machines is quite a valid use case for ordering. However in my experience, timestamp ordering is insufficient for state machines. Elements that cause state transitions might come in with the exact same timestamp, yet still have a necessary ordering. Especially given Beam's decision to have milliseconds timestamps this is possible, but even at microsecond or nanosecond precision this can happen at scale. To handle state machines you usually need some sort of FIFO ordering along with an ordered sources, such as Kafka, not timestamp ordering.
>>> >
>>> > Reuven
>>> >
>>> > On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz> wrote:
>>> >>
>>> >> Hi all,
>>> >>
>>> >> thanks everyone for this discussion. I think I have gathered enough
>>> >> feedback to be able to put down a proposition for changes, which I will
>>> >> do and send to this list for further discussion. There are still doubts
>>> >> remaining the non-determinism and it's relation to outputs stability vs.
>>> >> latency. But I will try to clarify all this in the design document.
>>> >>
>>> >> Thanks,
>>> >>
>>> >> Jan
>>> >>
>>> >> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>>> >> >> Someone from Flink might correct me if I'm wrong, but that's my
>>> >> >> current understanding.
>>> >> >
>>> >> > In essence your description of how exactly-once works in Flink is
>>> >> > correct. The general assumption in Flink is that pipelines must be
>>> >> > deterministic and thus produce idempotent writes in the case of
>>> >> > failures. However, that doesn't mean Beam sinks can't guarantee a bit
>>> >> > more with what Flink has to offer.
>>> >> >
>>> >> > Luke already mentioned the design discussions for @RequiresStableInput
>>> >> > which ensures idempotent writes for non-deterministic pipelines. This
>>> >> > is not part of the model but an optional Beam feature.
>>> >> >
>>> >> > We recently implemented support for @RequiresStableInput in the Flink
>>> >> > Runner. Reuven mentioned the Flink checkpoint confirmation, which
>>> >> > allows us to buffer (and checkpoint) processed data and only emit it
>>> >> > once a Flink checkpoint has completed.
>>> >> >
>>> >> > Cheers,
>>> >> > Max
>>> >> >
>>> >> > On 21.05.19 16:49, Jan Lukavský wrote:
>>> >> >> Hi,
>>> >> >>
>>> >> >> > Actually, I think it is a larger (open) question whether exactly
>>> >> >> once is guaranteed by the model or whether runners are allowed to
>>> >> >> relax that. I would think, however, that sources correctly
>>> >> >> implemented should be idempotent when run atop an exactly once
>>> >> >> infrastructure such as Flink of Dataflow.
>>> >> >>
>>> >> >> I would assume, that the model basically inherits guarantees of
>>> >> >> underlying infrastructure. Because Flink does not work as you
>>> >> >> described (atomic commit of inputs, state and outputs), but rather a
>>> >> >> checkpoint mark is flowing through the DAG much like watermark and on
>>> >> >> failures operators are restored and data reprocessed, it (IMHO)
>>> >> >> implies, that you have exactly once everywhere in the DAG *but*
>>> >> >> sinks. That is because sinks cannot be restored to previous state,
>>> >> >> instead sinks are supposed to be idempotent in order for the exactly
>>> >> >> once to really work (or at least be able to commit outputs on
>>> >> >> checkpoint in sink). That implies that if you don't have sink that is
>>> >> >> able to commit outputs atomically on checkpoint, the pipeline
>>> >> >> execution should be deterministic upon retries, otherwise shadow
>>> >> >> writes from failed paths of the pipeline might appear.
>>> >> >>
>>> >> >> Someone from Flink might correct me if I'm wrong, but that's my
>>> >> >> current understanding.
>>> >> >>
>>> >> >> > Sounds like we should make this clearer.
>>> >> >>
>>> >> >> I meant that you are right that we must not in any thoughts we are
>>> >> >> having forget that streams are by definition out-of-order. That is
>>> >> >> property that we cannot change. But - that doesn't limit us from
>>> >> >> creating operator that presents the data to UDF as if the stream was
>>> >> >> ideally sorted. It can do that by introducing latency, of course.
>>> >> >>
>>> >> >> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>>> >> >>> Reza: One could provide something like this as a utility class, but
>>> >> >>> one downside is that it is not scale invariant. It requires a tuning
>>> >> >>> parameter that, if to small, won't mitigate the problem, but if to
>>> >> >>> big, greatly increases latency. (Possibly one could define a dynamic
>>> >> >>> session-like window to solve this though...) It also might be harder
>>> >> >>> for runners that *can* cheaply present stuff in timestamp order to
>>> >> >>> optimize. (That and, in practice, our annotation-style process methods
>>> >> >>> don't lend themselves to easy composition.) I think it could work in
>>> >> >>> specific cases though.
>>> >> >>>
>>> >> >>> More inline below.
>>> >> >>>
>>> >> >>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz> wrote:
>>> >> >>>> Hi Robert,
>>> >> >>>>
>>> >> >>>> > Beam has an exactly-once model. If the data was consumed, state
>>> >> >>>> mutated, and outputs written downstream (these three are committed
>>> >> >>>> together atomically) it will not be replayed. That does not, of
>>> >> >>>> course,
>>> >> >>>> solve the non-determanism due to ordering (including the fact that two
>>> >> >>>> operations reading the same PCollection may view different ordering).
>>> >> >>>>
>>> >> >>>> I think what you describe is a property of a runner, not of the model,
>>> >> >>>> right? I think if I run my pipeline on Flink I will not get this
>>> >> >>>> atomicity, because although Flink uses also exactly-once model if
>>> >> >>>> might
>>> >> >>>> write outputs multiple times.
>>> >> >>> Actually, I think it is a larger (open) question whether exactly once
>>> >> >>> is guaranteed by the model or whether runners are allowed to relax
>>> >> >>> that. I would think, however, that sources correctly implemented
>>> >> >>> should be idempotent when run atop an exactly once infrastructure such
>>> >> >>> as Flink of Dataflow.
>>> >> >>>
>>> >> >>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
>>> >> >>>> received
>>> >> >>>> in a specific order? In the current model, it is not. Being able to
>>> >> >>>> read, handle, and produced out-of-order data, including late data,
>>> >> >>>> is a
>>> >> >>>> pretty fundamental property of distributed systems.
>>> >> >>>>
>>> >> >>>> Yes, absolutely. The argument here is not that Stateful ParDo should
>>> >> >>>> presume to receive elements in any order, but to _present_ it as
>>> >> >>>> such to
>>> >> >>>> the user @ProcessElement function.
>>> >> >>> Sounds like we should make this clearer.
>>> >> >>>
>>> >> >>>> > 2) Given that some operations are easier (or possibly only
>>> >> >>>> possible)
>>> >> >>>> to write when operating on ordered data, and that different runners
>>> >> >>>> may
>>> >> >>>> have (significantly) cheaper ways to provide this ordering than can be
>>> >> >>>> done by the user themselves, should we elevate this to a property of
>>> >> >>>> (Stateful?)DoFns that the runner can provide? I think a compelling
>>> >> >>>> argument can be made here that we should.
>>> >> >>>>
>>> >> >>>> +1
>>> >> >>>>
>>> >> >>>> Jan
>>> >> >>>>
>>> >> >>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>>> >> >>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz> wrote:
>>> >> >>>>>> > I don't see batch vs. streaming as part of the model. One
>>> >> >>>>>> can have
>>> >> >>>>>> microbatch, or even a runner that alternates between different
>>> >> >>>>>> modes.
>>> >> >>>>>>
>>> >> >>>>>> Although I understand motivation of this statement, this project
>>> >> >>>>>> name is
>>> >> >>>>>> "Apache Beam: An advanced unified programming model". What does the
>>> >> >>>>>> model unify, if "streaming vs. batch" is not part of the model?
>>> >> >>>>> What I mean is that streaming vs. batch is no longer part of the
>>> >> >>>>> model
>>> >> >>>>> (or ideally API), but pushed down to be a concern of the runner
>>> >> >>>>> (executor) of the pipeline.
>>> >> >>>>>
>>> >> >>>>>
>>> >> >>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz>
>>> >> >>>>> wrote:
>>> >> >>>>>> Hi Kenn,
>>> >> >>>>>>
>>> >> >>>>>> OK, so if we introduce annotation, we can have stateful ParDo
>>> >> >>>>>> with sorting, that would perfectly resolve my issues. I still
>>> >> >>>>>> have some doubts, though. Let me explain. The current behavior of
>>> >> >>>>>> stateful ParDo has the following properties:
>>> >> >>>>>>
>>> >> >>>>>> a) might fail in batch, although runs fine in streaming (that
>>> >> >>>>>> is due to the buffering, and unbounded lateness in batch, which
>>> >> >>>>>> was discussed back and forth in this thread)
>>> >> >>>>>>
>>> >> >>>>>> b) might be non deterministic (this is because the elements
>>> >> >>>>>> arrive at somewhat random order, and even if you do the operation
>>> >> >>>>>> "assign unique ID to elements" this might produce different
>>> >> >>>>>> results when run multiple times)
>>> >> >>>>> PCollections are *explicitly* unordered. Any operations that
>>> >> >>>>> assume or
>>> >> >>>>> depend on a specific ordering for correctness (or determinism) must
>>> >> >>>>> provide that ordering themselves (i.e. tolerate "arbitrary shuffling
>>> >> >>>>> of inputs"). As you point out, that may be very expensive if you have
>>> >> >>>>> very hot keys with very large (unbounded) timestamp skew.
>>> >> >>>>>
>>> >> >>>>> StatefulDoFns are low-level operations that should be used with care;
>>> >> >>>>> the simpler windowing model gives determinism in the face of
>>> >> >>>>> unordered
>>> >> >>>>> data (though late data and non-end-of-window triggering introduces
>>> >> >>>>> some of the non-determanism back in).
>>> >> >>>>>
>>> >> >>>>>> What worries me most is the property b), because it seems to me
>>> >> >>>>>> to have serious consequences - not only that if you run twice
>>> >> >>>>>> batch pipeline you would get different results, but even on
>>> >> >>>>>> streaming, when pipeline fails and gets restarted from
>>> >> >>>>>> checkpoint, produced output might differ from the previous run
>>> >> >>>>>> and data from the first run might have already been persisted
>>> >> >>>>>> into sink. That would create somewhat messy outputs.
>>> >> >>>>> Beam has an exactly-once model. If the data was consumed, state
>>> >> >>>>> mutated, and outputs written downstream (these three are committed
>>> >> >>>>> together atomically) it will not be replayed. That does not, of
>>> >> >>>>> course, solve the non-determanism due to ordering (including the fact
>>> >> >>>>> that two operations reading the same PCollection may view different
>>> >> >>>>> ordering).
>>> >> >>>>>
>>> >> >>>>>> These two properties makes me think that the current
>>> >> >>>>>> implementation is more of a _special case_ than the general one.
>>> >> >>>>>> The general one would be that your state doesn't have the
>>> >> >>>>>> properties to be able to tolerate buffering problems and/or
>>> >> >>>>>> non-determinism. Which is the case where you need sorting in both
>>> >> >>>>>> streaming and batch to be part of the model.
>>> >> >>>>>>
>>> >> >>>>>> Let me point out one more analogy - that is merging vs.
>>> >> >>>>>> non-merging windows. The general case (merging windows) implies
>>> >> >>>>>> sorting by timestamp in both batch case (explicit) and streaming
>>> >> >>>>>> (buffering). The special case (non-merging windows) doesn't rely
>>> >> >>>>>> on any timestamp ordering, so the sorting and buffering can be
>>> >> >>>>>> dropped. The underlying root cause of this is the same for both
>>> >> >>>>>> stateful ParDo and windowing (essentially, assigning window
>>> >> >>>>>> labels is a stateful operation when windowing function is merging).
>>> >> >>>>>>
>>> >> >>>>>> The reason for the current behavior of stateful ParDo seems to be
>>> >> >>>>>> performance, but is it right to abandon correctness in favor of
>>> >> >>>>>> performance? Wouldn't it be more consistent to have the default
>>> >> >>>>>> behavior prefer correctness and when you have the specific
>>> >> >>>>>> conditions of state function having special properties, then you
>>> >> >>>>>> can annotate your DoFn (with something like
>>> >> >>>>>> @TimeOrderingAgnostic), which would yield a better performance in
>>> >> >>>>>> that case?
>>> >> >>>>> There are two separable questions here.
>>> >> >>>>>
>>> >> >>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are received
>>> >> >>>>> in a specific order? In the current model, it is not. Being able to
>>> >> >>>>> read, handle, and produced out-of-order data, including late data, is
>>> >> >>>>> a pretty fundamental property of distributed systems.
>>> >> >>>>>
>>> >> >>>>> 2) Given that some operations are easier (or possibly only possible)
>>> >> >>>>> to write when operating on ordered data, and that different runners
>>> >> >>>>> may have (significantly) cheaper ways to provide this ordering than
>>> >> >>>>> can be done by the user themselves, should we elevate this to a
>>> >> >>>>> property of (Stateful?)DoFns that the runner can provide? I think a
>>> >> >>>>> compelling argument can be made here that we should.
>>> >> >>>>>
>>> >> >>>>> - Robert
>>> >> >>>>>
>>> >> >>>>>
>>> >> >>>>>
>>> >> >>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>>> >> >>>>>>
>>> >> >>>>>> Thanks for the nice small example of a calculation that depends
>>> >> >>>>>> on order. You are right that many state machines have this
>>> >> >>>>>> property. I agree w/ you and Luke that it is convenient for batch
>>> >> >>>>>> processing to sort by event timestamp before running a stateful
>>> >> >>>>>> ParDo. In streaming you could also implement "sort by event
>>> >> >>>>>> timestamp" by buffering until you know all earlier data will be
>>> >> >>>>>> dropped - a slack buffer up to allowed lateness.
>>> >> >>>>>>
>>> >> >>>>>> I do not think that it is OK to sort in batch and not in
>>> >> >>>>>> streaming. Many state machines diverge very rapidly when things
>>> >> >>>>>> are out of order. So each runner if they see the
>>> >> >>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
>>> >> >>>>>> sorted data (by some mix of buffering and dropping), or to reject
>>> >> >>>>>> the pipeline as unsupported.
>>> >> >>>>>>
>>> >> >>>>>> And also want to say that this is not the default case - many
>>> >> >>>>>> uses of state & timers in ParDo yield different results at the
>>> >> >>>>>> element level, but the results are equivalent at in the big
>>> >> >>>>>> picture. Such as the example of "assign a unique sequence number
>>> >> >>>>>> to each element" or "group into batches" it doesn't matter
>>> >> >>>>>> exactly what the result is, only that it meets the spec. And
>>> >> >>>>>> other cases like user funnels are monotonic enough that you also
>>> >> >>>>>> don't actually need sorting.
>>> >> >>>>>>
>>> >> >>>>>> Kenn
>>> >> >>>>>>
>>> >> >>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz>
>>> >> >>>>>> wrote:
>>> >> >>>>>>> Yes, the problem will arise probably mostly when you have not
>>> >> >>>>>>> well distributed keys (or too few keys). I'm really not sure if
>>> >> >>>>>>> a pure GBK with a trigger can solve this - it might help to have
>>> >> >>>>>>> data driven trigger. There would still be some doubts, though.
>>> >> >>>>>>> The main question is still here - people say, that sorting by
>>> >> >>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
>>> >> >>>>>>> I don't really see why - the sorting is very probably already
>>> >> >>>>>>> there. And if not (hash grouping instead of sorted grouping),
>>> >> >>>>>>> then the sorting would affect only user defined StatefulParDos.
>>> >> >>>>>>>
>>> >> >>>>>>> This would suggest that the best way out of this would be really
>>> >> >>>>>>> to add annotation, so that the author of the pipeline can decide.
>>> >> >>>>>>>
>>> >> >>>>>>> If that would be acceptable I think I can try to prepare some
>>> >> >>>>>>> basic functionality, but I'm not sure, if I would be able to
>>> >> >>>>>>> cover all runners / sdks.
>>> >> >>>>>>>
>>> >> >>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>>> >> >>>>>>>
>>> >> >>>>>>> It is read all per key and window and not just read all (this
>>> >> >>>>>>> still won't scale with hot keys in the global window). The GBK
>>> >> >>>>>>> preceding the StatefulParDo will guarantee that you are
>>> >> >>>>>>> processing all the values for a specific key and window at any
>>> >> >>>>>>> given time. Is there a specific window/trigger that is missing
>>> >> >>>>>>> that you feel would remove the need for you to use StatefulParDo?
>>> >> >>>>>>>
>>> >> >>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
>>> >> >>>>>>> wrote:
>>> >> >>>>>>>> Hi Lukasz,
>>> >> >>>>>>>>
>>> >> >>>>>>>>> Today, if you must have a strict order, you must guarantee
>>> >> >>>>>>>>> that your StatefulParDo implements the necessary "buffering &
>>> >> >>>>>>>>> sorting" into state.
>>> >> >>>>>>>> Yes, no problem with that. But this whole discussion started,
>>> >> >>>>>>>> because *this doesn't work on batch*. You simply cannot first
>>> >> >>>>>>>> read everything from distributed storage and then buffer it all
>>> >> >>>>>>>> into memory, just to read it again, but sorted. That will not
>>> >> >>>>>>>> work. And even if it would, it would be a terrible waste of
>>> >> >>>>>>>> resources.
>>> >> >>>>>>>>
>>> >> >>>>>>>> Jan
>>> >> >>>>>>>>
>>> >> >>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>>> >> >>>>>>>>
>>> >> >>>>>>>>
>>> >> >>>>>>>>
>>> >> >>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
>>> >> >>>>>>>> wrote:
>>> >> >>>>>>>>> This discussion brings many really interesting questions for
>>> >> >>>>>>>>> me. :-)
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> > I don't see batch vs. streaming as part of the model. One
>>> >> >>>>>>>>> can have
>>> >> >>>>>>>>> microbatch, or even a runner that alternates between different
>>> >> >>>>>>>>> modes.
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> Although I understand motivation of this statement, this
>>> >> >>>>>>>>> project name is
>>> >> >>>>>>>>> "Apache Beam: An advanced unified programming model". What
>>> >> >>>>>>>>> does the
>>> >> >>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
>>> >> >>>>>>>>> are
>>> >> >>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
>>> >> >>>>>>>>> All these
>>> >> >>>>>>>>> define several runtime parameters, which in turn define how
>>> >> >>>>>>>>> well/badly
>>> >> >>>>>>>>> will the pipeline perform and how many resources might be
>>> >> >>>>>>>>> needed. From
>>> >> >>>>>>>>> my point of view, pure streaming should be the most resource
>>> >> >>>>>>>>> demanding
>>> >> >>>>>>>>> (if not, why bother with batch? why not run everything in
>>> >> >>>>>>>>> streaming
>>> >> >>>>>>>>> only? what will there remain to "unify"?).
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> > Fortunately, for batch, only the state for a single key
>>> >> >>>>>>>>> needs to be
>>> >> >>>>>>>>> preserved at a time, rather than the state for all keys across
>>> >> >>>>>>>>> the range
>>> >> >>>>>>>>> of skew. Of course if you have few or hot keys, one can still
>>> >> >>>>>>>>> have
>>> >> >>>>>>>>> issues (and this is not specific to StatefulDoFns).
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> Yes, but here is still the presumption that my stateful DoFn can
>>> >> >>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
>>> >> >>>>>>>>> case in
>>> >> >>>>>>>>> more detail.
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
>>> >> >>>>>>>>> some key for
>>> >> >>>>>>>>> each element, which is irrelevant for the demonstration). Your
>>> >> >>>>>>>>> task is
>>> >> >>>>>>>>> to calculate in running global window the actual number of
>>> >> >>>>>>>>> changes
>>> >> >>>>>>>>> between state 0 and state 1 and vice versa. When the state
>>> >> >>>>>>>>> doesn't
>>> >> >>>>>>>>> change, you don't calculate anything. If input (for given key)
>>> >> >>>>>>>>> would be
>>> >> >>>>>>>>> (tN denotes timestamp N):
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t1: 1
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t2: 0
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t3: 0
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t4: 1
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t5: 1
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t6: 0
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> then the output should yield (supposing that default state is
>>> >> >>>>>>>>> zero):
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t1: (one: 1, zero: 0)
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t2: (one: 1, zero: 1)
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t3: (one: 1, zero: 1)
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t4: (one: 2, zero: 1)
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t5: (one: 2, zero: 1)
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> t6: (one: 2, zero: 2)
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> How would you implement this in current Beam semantics?
>>> >> >>>>>>>> I think your saying here that I know that my input is ordered
>>> >> >>>>>>>> in a specific way and since I assume the order when writing my
>>> >> >>>>>>>> pipeline I can perform this optimization. But there is nothing
>>> >> >>>>>>>> preventing a runner from noticing that your processing in the
>>> >> >>>>>>>> global window with a specific type of trigger and re-ordering
>>> >> >>>>>>>> your inputs/processing to get better performance (since you
>>> >> >>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>>> >> >>>>>>>> streaming for the GlobalWindow).
>>> >> >>>>>>>>
>>> >> >>>>>>>> Today, if you must have a strict order, you must guarantee that
>>> >> >>>>>>>> your StatefulParDo implements the necessary "buffering &
>>> >> >>>>>>>> sorting" into state. I can see why you would want an annotation
>>> >> >>>>>>>> that says I must have timestamp ordered elements, since it
>>> >> >>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
>>> >> >>>>>>>> is a low-level function, it really is the "here you go and do
>>> >> >>>>>>>> whatever you need to but here be dragons" function while
>>> >> >>>>>>>> windowing and triggering is meant to keep many people from
>>> >> >>>>>>>> writing StatefulParDo in the first place.
>>> >> >>>>>>>>
>>> >> >>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
>>> >> >>>>>>>>> are likely to
>>> >> >>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>>> >> >>>>>>>>> falls
>>> >> >>>>>>>>> behind in streaming mode as well.
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> But the worst case is defined by input of size (available
>>> >> >>>>>>>>> resources +
>>> >> >>>>>>>>> single byte) -> pipeline fail. Although it could have
>>> >> >>>>>>>>> finished, given
>>> >> >>>>>>>>> the right conditions.
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> > This might be reasonable, implemented by default by
>>> >> >>>>>>>>> buffering
>>> >> >>>>>>>>> everything and releasing elements as the watermark (+lateness)
>>> >> >>>>>>>>> advances,
>>> >> >>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
>>> >> >>>>>>>>> reason
>>> >> >>>>>>>>> about) code.
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> Sure, the pipeline will be less efficient, because it would
>>> >> >>>>>>>>> have to
>>> >> >>>>>>>>> buffer and sort the inputs. But at least it will produce
>>> >> >>>>>>>>> correct results
>>> >> >>>>>>>>> in cases where updates to state are order-sensitive.
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
>>> >> >>>>>>>>> (key, values):
>>> >> >>>>>>>>> [(key, value) for value in values])?
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
>>> >> >>>>>>>>> The trigger
>>> >> >>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
>>> >> >>>>>>>>> element
>>> >> >>>>>>>>> with lowest timestamp in the buffer. Although this could be
>>> >> >>>>>>>>> somehow
>>> >> >>>>>>>>> emulated by fixed trigger each X millis.
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> > Or is the underlying desire just to be able to hint to
>>> >> >>>>>>>>> the runner
>>> >> >>>>>>>>> that the code may perform better (e.g. require less resources)
>>> >> >>>>>>>>> as skew
>>> >> >>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> No, the sorting would have to be done in streaming case as
>>> >> >>>>>>>>> well. That is
>>> >> >>>>>>>>> an imperative of the unified model. I think it is possible to
>>> >> >>>>>>>>> sort by
>>> >> >>>>>>>>> timestamp only in batch case (and do it for *all* batch
>>> >> >>>>>>>>> stateful pardos
>>> >> >>>>>>>>> without annotation), or introduce annotation, but then make
>>> >> >>>>>>>>> the same
>>> >> >>>>>>>>> guarantees for streaming case as well.
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> Jan
>>> >> >>>>>>>>>
>>> >> >>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>>> >> >>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>>> >> >>>>>>>>>> <je...@seznam.cz> wrote:
>>> >> >>>>>>>>>>> Hi Robert,
>>> >> >>>>>>>>>>>
>>> >> >>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
>>> >> >>>>>>>>>>> guarantees
>>> >> >>>>>>>>>>> of ordering are given in either mode, there is *implicit*
>>> >> >>>>>>>>>>> ordering in
>>> >> >>>>>>>>>>> streaming case that is due to nature of the processing - the
>>> >> >>>>>>>>>>> difference
>>> >> >>>>>>>>>>> between watermark and timestamp of elements flowing through
>>> >> >>>>>>>>>>> the pipeline
>>> >> >>>>>>>>>>> are generally low (too high difference leads to the
>>> >> >>>>>>>>>>> overbuffering
>>> >> >>>>>>>>>>> problem), but there is no such bound on batch.
>>> >> >>>>>>>>>> Fortunately, for batch, only the state for a single key needs
>>> >> >>>>>>>>>> to be
>>> >> >>>>>>>>>> preserved at a time, rather than the state for all keys
>>> >> >>>>>>>>>> across the
>>> >> >>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
>>> >> >>>>>>>>>> still
>>> >> >>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>>> >> >>>>>>>>>>
>>> >> >>>>>>>>>>> As a result, I see a few possible solutions:
>>> >> >>>>>>>>>>>
>>> >> >>>>>>>>>>> - the best and most natural seems to be extension of
>>> >> >>>>>>>>>>> the model, so
>>> >> >>>>>>>>>>> that it defines batch as not only "streaming pipeline
>>> >> >>>>>>>>>>> executed in batch
>>> >> >>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>>> >> >>>>>>>>>>> characteristics as
>>> >> >>>>>>>>>>> in streaming case, executed in batch fashion", I really
>>> >> >>>>>>>>>>> don't think that
>>> >> >>>>>>>>>>> there are any conflicts with the current model, or that this
>>> >> >>>>>>>>>>> could
>>> >> >>>>>>>>>>> affect performance, because the required sorting (as pointed by
>>> >> >>>>>>>>>>> Aljoscha) is very probably already done during translation
>>> >> >>>>>>>>>>> of stateful
>>> >> >>>>>>>>>>> pardos. Also note that this definition only affects user
>>> >> >>>>>>>>>>> defined
>>> >> >>>>>>>>>>> stateful pardos
>>> >> >>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
>>> >> >>>>>>>>>> have
>>> >> >>>>>>>>>> microbatch, or even a runner that alternates between
>>> >> >>>>>>>>>> different modes.
>>> >> >>>>>>>>>> The model describes what the valid outputs are given a
>>> >> >>>>>>>>>> (sometimes
>>> >> >>>>>>>>>> partial) set of inputs. It becomes really hard to define
>>> >> >>>>>>>>>> things like
>>> >> >>>>>>>>>> "as good runtime characteristics." Once you allow any
>>> >> >>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
>>> >> >>>>>>>>>> (and
>>> >> >>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>>> >> >>>>>>>>>> out-of-orderedness.
>>> >> >>>>>>>>>>
>>> >> >>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
>>> >> >>>>>>>>>> likely to
>>> >> >>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>>> >> >>>>>>>>>> falls
>>> >> >>>>>>>>>> behind in streaming mode as well.
>>> >> >>>>>>>>>>
>>> >> >>>>>>>>>>> - another option would be to introduce annotation for
>>> >> >>>>>>>>>>> DoFns (e.g.
>>> >> >>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
>>> >> >>>>>>>>>>> the sorting
>>> >> >>>>>>>>>>> in batch case - but - this extension would have to ensure
>>> >> >>>>>>>>>>> the sorting in
>>> >> >>>>>>>>>>> streaming mode also - it would require definition of allowed
>>> >> >>>>>>>>>>> lateness,
>>> >> >>>>>>>>>>> and triggger (essentially similar to window)
>>> >> >>>>>>>>>> This might be reasonable, implemented by default by buffering
>>> >> >>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>>> >> >>>>>>>>>> advances, but would likely lead to inefficient (though
>>> >> >>>>>>>>>> *maybe* easier
>>> >> >>>>>>>>>> to reason about) code. Not sure about the semantics of
>>> >> >>>>>>>>>> triggering
>>> >> >>>>>>>>>> here, especially data-driven triggers. Would it be roughly
>>> >> >>>>>>>>>> equivalent
>>> >> >>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
>>> >> >>>>>>>>>> value in
>>> >> >>>>>>>>>> values])?
>>> >> >>>>>>>>>>
>>> >> >>>>>>>>>> Or is the underlying desire just to be able to hint to the
>>> >> >>>>>>>>>> runner that
>>> >> >>>>>>>>>> the code may perform better (e.g. require less resources) as
>>> >> >>>>>>>>>> skew is
>>> >> >>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>>> >> >>>>>>>>>>
>>> >> >>>>>>>>>>> - last option would be to introduce these "higher order
>>> >> >>>>>>>>>>> guarantees" in
>>> >> >>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
>>> >> >>>>>>>>>>> worst
>>> >> >>>>>>>>>>> option to me
>>> >> >>>>>>>>>>>
>>> >> >>>>>>>>>>> I see the first two options quite equally good, although the
>>> >> >>>>>>>>>>> letter one
>>> >> >>>>>>>>>>> is probably more time consuming to implement. But it would
>>> >> >>>>>>>>>>> bring
>>> >> >>>>>>>>>>> additional feature to streaming case as well.
>>> >> >>>>>>>>>>>
>>> >> >>>>>>>>>>> Thanks for any thoughts.
>>> >> >>>>>>>>>>>
>>> >> >>>>>>>>>>> Jan
>>> >> >>>>>>>>>>>
>>> >> >>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>>> >> >>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>>> >> >>>>>>>>>>>> <je...@seznam.cz> wrote:
>>> >> >>>>>>>>>>>>> Hi Reuven,
>>> >> >>>>>>>>>>>>>
>>> >> >>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
>>> >> >>>>>>>>>>>>>> runners.
>>> >> >>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
>>> >> >>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
>>> >> >>>>>>>>>>>>> of elements. That basically (practically) can work only
>>> >> >>>>>>>>>>>>> for cases, where the order of input elements doesn't
>>> >> >>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
>>> >> >>>>>>>>>>>>> time you have a state machine involved, then the ordering
>>> >> >>>>>>>>>>>>> of elements would matter.
>>> >> >>>>>>>>>>>> No guarantees on order are provided in *either* streaming
>>> >> >>>>>>>>>>>> or batch
>>> >> >>>>>>>>>>>> mode by the model. However, it is the case that in order to
>>> >> >>>>>>>>>>>> make
>>> >> >>>>>>>>>>>> forward progress most streaming runners attempt to limit
>>> >> >>>>>>>>>>>> the amount of
>>> >> >>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
>>> >> >>>>>>>>>>>> processing
>>> >> >>>>>>>>>>>> time) to make forward progress, which in turn could help
>>> >> >>>>>>>>>>>> cap the
>>> >> >>>>>>>>>>>> amount of state that must be held concurrently, whereas a
>>> >> >>>>>>>>>>>> batch runner
>>> >> >>>>>>>>>>>> may not allow any state to be safely discarded until the whole
>>> >> >>>>>>>>>>>> timeline from infinite past to infinite future has been
>>> >> >>>>>>>>>>>> observed.
>>> >> >>>>>>>>>>>>
>>> >> >>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
>>> >> >>>>>>>>>>>> batch" in batch mode.
>>> >> >>>>>>>>>>>>
>>> >> >>>>>>>>>>>>
>>> >> >>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>>> >> >>>>>>>>>>>> <mx...@apache.org> wrote:
>>> >> >>>>>>>>>>>>
>>> >> >>>>>>>>>>>>>> batch semantics and streaming semantics differs only
>>> >> >>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
>>> >> >>>>>>>>>>>>>> batch and cannot on stream
>>> >> >>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
>>> >> >>>>>>>>>>>>> trigger. You
>>> >> >>>>>>>>>>>>> could define additional triggers that do early firings.
>>> >> >>>>>>>>>>>>> And you could
>>> >> >>>>>>>>>>>>> even trigger the global window by advancing the watermark
>>> >> >>>>>>>>>>>>> to +inf.
>>> >> >>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
>>> >> >>>>>>>>>>>> default
>>> >> >>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
>>> >> >>>>>>>>>>>> is more
>>> >> >>>>>>>>>>>> likely to be user error than an actual desire to have no
>>> >> >>>>>>>>>>>> output until
>>> >> >>>>>>>>>>>> drain. But it's semantically valid in the model.
Re: Definition of Unified model
Posted by Kenneth Knowles <ke...@apache.org>.
I strongly prefer explicit sequence metadata over FIFO requirements,
because:
- FIFO is complex to specify: for example Dataflow has "per stage
key-to-key" FIFO today, but it is not guaranteed to remain so (plus "stage"
is not a portable concept, nor even guaranteed to remain a Dataflow concept)
- complex specifications are by definition poor usability (if necessary,
then it is what it is)
- overly restricts the runner, reduces parallelism, for example any
non-stateful ParDo has per-element parallelism, not per "key"
- another perspective on that: FIFO makes everyone pay rather than just
the transform that requires exactly sequencing
- previous implementation details like reshuffles become part of the model
- I'm not even convinced the use cases involved are addressed by some
careful FIFO restrictions; many sinks re-key and they would all have to
become aware of how keying of a sequence of "stages" affects the end-to-end
FIFO
A noop becoming a non-noop is essentially the mathematical definition of
moving from higher-level to lower-level abstraction.
So this strikes at the core question of what level of abstraction Beam aims
to represent. Lower-level means there are fewer possible implementations
and it is more tied to the underlying architecture, and anything not
near-exact match pays a huge penalty. Higher-level means there are more
implementations possible with different tradeoffs, though they may all pay
a minor penalty.
I could be convinced to change my mind, but it needs some extensive design,
examples, etc. I think it is probably about the most consequential design
decision in the whole Beam model, around the same level as the decision to
use ParDo and GBK as the primitives IMO.
Kenn
On Thu, May 23, 2019 at 10:17 AM Reuven Lax <re...@google.com> wrote:
> Not really. I'm suggesting that some variant of FIFO ordering is
> necessary, which requires either runners natively support FIFO ordering or
> transforms adding some extra sequence number to each record to sort by.
>
> I still think your proposal is very useful by the way. I'm merely pointing
> out that to solve the state-machine problem we probably need something more.
>
> Reuven
>
> On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je...@seznam.cz> wrote:
>
>> Hi,
>> yes. It seems that ordering by user supplied UDF makes sense and I will
>> update the design proposal accordingly.
>> Would that solve the issues you mention?
>> Jan
>> ---------- Původní e-mail ----------
>> Od: Reuven Lax <re...@google.com>
>> Komu: dev <de...@beam.apache.org>
>> Datum: 23. 5. 2019 18:44:38
>> Předmět: Re: Definition of Unified model
>>
>> I'm simply saying that timestamp ordering is insufficient for state
>> machines. I wasn't proposing Kafka as a solution - that was simply an
>> example of how people solve this problem in other scenarios.
>>
>> BTW another example of ordering: Imagine today that you have a triggered
>> Sum aggregation writing out to a key-value sink. In theory we provide no
>> ordering, so the sink might write the triggered sums in the wrong order,
>> ending up with an incorrect value in the sink. In this case you probably
>> want values ordered by trigger pane index.
>>
>> Reuven
>>
>> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz> wrote:
>>
>> Hi Reuven,
>> I share the view point of Robert. I think the isuue you refer to is not
>> in reality related to timestamps, but to the fact, that ordering of events
>> in time is observer dependent (either caused by relativity, or time skew,
>> essentially this has the same consequences). And the resolution in fact
>> isn't Kafka, but generally an authoritative observer, that tells you "I saw
>> the events in this order". And you either have one (and have the outcome of
>> his observation persisted in the data - e.g. as offset in Kafka partition),
>> then you should be able to use it (maybe that suggests afterall that
>> sorting by some user supplied UDF might make sense), or do not have it, and
>> then any interpretation of the data seems to be equally valid. Although
>> determinism is fine, of course.
>> Jan
>> ---------- Původní e-mail ----------
>> Od: Reuven Lax <re...@google.com>
>> Komu: dev <de...@beam.apache.org>
>> Datum: 23. 5. 2019 17:39:12
>> Předmět: Re: Definition of Unified model
>>
>> So an example would be elements of type "startUserSession" and
>> "endUserSession" (website sessions, not Beam sessions). Logically you may
>> need to process them in the correct order if you have any sort of
>> state-machine logic. However timestamp ordering is never guaranteed to
>> match the logical ordering. Not only might you have several elements with
>> the same timestamp, but in reality time skew across backend servers can
>> cause the events to have timestamps in reverse order of the actual
>> causality order.
>>
>> People do solve this problem today though. Publish the events to Kafka,
>> making sure that events for the same user end up in the same Kafka
>> partition. This ensures that the events appear in the Kafka partitions in
>> causality order, even if the timestamp order doesn't match. The your Kafka
>> subscriber simply process the elements in each partition in order.
>>
>> I think the ability to impose FIFO causality ordering is what's needed
>> for any state-machine work. Timestamp ordering has advantages (though often
>> I think the advantage is in state), but does not solve this problem.
>>
>> Reuven
>>
>> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <ro...@google.com>
>> wrote:
>>
>> Good point.
>>
>> The "implementation-specific" way I would do this is
>> window-by-instant, followed by a DoFn that gets all the elements with
>> the same timestamp and sorts/acts accordingly, but this counts on the
>> runner producing windows in timestamp order (likely?) and also the
>> subsequent DoFn getting them in this order (also likely, due to
>> fusion).
>>
>> One could make the argument that, though it does not provide
>> deterministic behavior, getting elements of the same timestamp in
>> different orders should produce equally valid interpretations of the
>> data. (After all, due to relatively, timestamps are not technically
>> well ordered across space.) I can see how data-dependent tiebreakers
>> could be useful, or promises of preservation of order between
>> operations.
>>
>> - Robert
>>
>> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com> wrote:
>> >
>> > So Jan's example of state machines is quite a valid use case for
>> ordering. However in my experience, timestamp ordering is insufficient for
>> state machines. Elements that cause state transitions might come in with
>> the exact same timestamp, yet still have a necessary ordering. Especially
>> given Beam's decision to have milliseconds timestamps this is possible, but
>> even at microsecond or nanosecond precision this can happen at scale. To
>> handle state machines you usually need some sort of FIFO ordering along
>> with an ordered sources, such as Kafka, not timestamp ordering.
>> >
>> > Reuven
>> >
>> > On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz> wrote:
>> >>
>> >> Hi all,
>> >>
>> >> thanks everyone for this discussion. I think I have gathered enough
>> >> feedback to be able to put down a proposition for changes, which I will
>> >> do and send to this list for further discussion. There are still doubts
>> >> remaining the non-determinism and it's relation to outputs stability
>> vs.
>> >> latency. But I will try to clarify all this in the design document.
>> >>
>> >> Thanks,
>> >>
>> >> Jan
>> >>
>> >> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>> >> >> Someone from Flink might correct me if I'm wrong, but that's my
>> >> >> current understanding.
>> >> >
>> >> > In essence your description of how exactly-once works in Flink is
>> >> > correct. The general assumption in Flink is that pipelines must be
>> >> > deterministic and thus produce idempotent writes in the case of
>> >> > failures. However, that doesn't mean Beam sinks can't guarantee a bit
>> >> > more with what Flink has to offer.
>> >> >
>> >> > Luke already mentioned the design discussions for
>> @RequiresStableInput
>> >> > which ensures idempotent writes for non-deterministic pipelines. This
>> >> > is not part of the model but an optional Beam feature.
>> >> >
>> >> > We recently implemented support for @RequiresStableInput in the Flink
>> >> > Runner. Reuven mentioned the Flink checkpoint confirmation, which
>> >> > allows us to buffer (and checkpoint) processed data and only emit it
>> >> > once a Flink checkpoint has completed.
>> >> >
>> >> > Cheers,
>> >> > Max
>> >> >
>> >> > On 21.05.19 16:49, Jan Lukavský wrote:
>> >> >> Hi,
>> >> >>
>> >> >> > Actually, I think it is a larger (open) question whether exactly
>> >> >> once is guaranteed by the model or whether runners are allowed to
>> >> >> relax that. I would think, however, that sources correctly
>> >> >> implemented should be idempotent when run atop an exactly once
>> >> >> infrastructure such as Flink of Dataflow.
>> >> >>
>> >> >> I would assume, that the model basically inherits guarantees of
>> >> >> underlying infrastructure. Because Flink does not work as you
>> >> >> described (atomic commit of inputs, state and outputs), but rather a
>> >> >> checkpoint mark is flowing through the DAG much like watermark and
>> on
>> >> >> failures operators are restored and data reprocessed, it (IMHO)
>> >> >> implies, that you have exactly once everywhere in the DAG *but*
>> >> >> sinks. That is because sinks cannot be restored to previous state,
>> >> >> instead sinks are supposed to be idempotent in order for the exactly
>> >> >> once to really work (or at least be able to commit outputs on
>> >> >> checkpoint in sink). That implies that if you don't have sink that
>> is
>> >> >> able to commit outputs atomically on checkpoint, the pipeline
>> >> >> execution should be deterministic upon retries, otherwise shadow
>> >> >> writes from failed paths of the pipeline might appear.
>> >> >>
>> >> >> Someone from Flink might correct me if I'm wrong, but that's my
>> >> >> current understanding.
>> >> >>
>> >> >> > Sounds like we should make this clearer.
>> >> >>
>> >> >> I meant that you are right that we must not in any thoughts we are
>> >> >> having forget that streams are by definition out-of-order. That is
>> >> >> property that we cannot change. But - that doesn't limit us from
>> >> >> creating operator that presents the data to UDF as if the stream was
>> >> >> ideally sorted. It can do that by introducing latency, of course.
>> >> >>
>> >> >> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>> >> >>> Reza: One could provide something like this as a utility class, but
>> >> >>> one downside is that it is not scale invariant. It requires a
>> tuning
>> >> >>> parameter that, if to small, won't mitigate the problem, but if to
>> >> >>> big, greatly increases latency. (Possibly one could define a
>> dynamic
>> >> >>> session-like window to solve this though...) It also might be
>> harder
>> >> >>> for runners that *can* cheaply present stuff in timestamp order to
>> >> >>> optimize. (That and, in practice, our annotation-style process
>> methods
>> >> >>> don't lend themselves to easy composition.) I think it could work
>> in
>> >> >>> specific cases though.
>> >> >>>
>> >> >>> More inline below.
>> >> >>>
>> >> >>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz>
>> wrote:
>> >> >>>> Hi Robert,
>> >> >>>>
>> >> >>>> > Beam has an exactly-once model. If the data was consumed,
>> state
>> >> >>>> mutated, and outputs written downstream (these three are committed
>> >> >>>> together atomically) it will not be replayed. That does not, of
>> >> >>>> course,
>> >> >>>> solve the non-determanism due to ordering (including the fact
>> that two
>> >> >>>> operations reading the same PCollection may view different
>> ordering).
>> >> >>>>
>> >> >>>> I think what you describe is a property of a runner, not of the
>> model,
>> >> >>>> right? I think if I run my pipeline on Flink I will not get this
>> >> >>>> atomicity, because although Flink uses also exactly-once model if
>> >> >>>> might
>> >> >>>> write outputs multiple times.
>> >> >>> Actually, I think it is a larger (open) question whether exactly
>> once
>> >> >>> is guaranteed by the model or whether runners are allowed to relax
>> >> >>> that. I would think, however, that sources correctly implemented
>> >> >>> should be idempotent when run atop an exactly once infrastructure
>> such
>> >> >>> as Flink of Dataflow.
>> >> >>>
>> >> >>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
>> >> >>>> received
>> >> >>>> in a specific order? In the current model, it is not. Being able
>> to
>> >> >>>> read, handle, and produced out-of-order data, including late data,
>> >> >>>> is a
>> >> >>>> pretty fundamental property of distributed systems.
>> >> >>>>
>> >> >>>> Yes, absolutely. The argument here is not that Stateful ParDo
>> should
>> >> >>>> presume to receive elements in any order, but to _present_ it as
>> >> >>>> such to
>> >> >>>> the user @ProcessElement function.
>> >> >>> Sounds like we should make this clearer.
>> >> >>>
>> >> >>>> > 2) Given that some operations are easier (or possibly only
>> >> >>>> possible)
>> >> >>>> to write when operating on ordered data, and that different
>> runners
>> >> >>>> may
>> >> >>>> have (significantly) cheaper ways to provide this ordering than
>> can be
>> >> >>>> done by the user themselves, should we elevate this to a property
>> of
>> >> >>>> (Stateful?)DoFns that the runner can provide? I think a compelling
>> >> >>>> argument can be made here that we should.
>> >> >>>>
>> >> >>>> +1
>> >> >>>>
>> >> >>>> Jan
>> >> >>>>
>> >> >>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>> >> >>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz>
>> wrote:
>> >> >>>>>> > I don't see batch vs. streaming as part of the model. One
>> >> >>>>>> can have
>> >> >>>>>> microbatch, or even a runner that alternates between different
>> >> >>>>>> modes.
>> >> >>>>>>
>> >> >>>>>> Although I understand motivation of this statement, this project
>> >> >>>>>> name is
>> >> >>>>>> "Apache Beam: An advanced unified programming model". What does
>> the
>> >> >>>>>> model unify, if "streaming vs. batch" is not part of the model?
>> >> >>>>> What I mean is that streaming vs. batch is no longer part of the
>> >> >>>>> model
>> >> >>>>> (or ideally API), but pushed down to be a concern of the runner
>> >> >>>>> (executor) of the pipeline.
>> >> >>>>>
>> >> >>>>>
>> >> >>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz>
>> >> >>>>> wrote:
>> >> >>>>>> Hi Kenn,
>> >> >>>>>>
>> >> >>>>>> OK, so if we introduce annotation, we can have stateful ParDo
>> >> >>>>>> with sorting, that would perfectly resolve my issues. I still
>> >> >>>>>> have some doubts, though. Let me explain. The current behavior
>> of
>> >> >>>>>> stateful ParDo has the following properties:
>> >> >>>>>>
>> >> >>>>>> a) might fail in batch, although runs fine in streaming (that
>> >> >>>>>> is due to the buffering, and unbounded lateness in batch, which
>> >> >>>>>> was discussed back and forth in this thread)
>> >> >>>>>>
>> >> >>>>>> b) might be non deterministic (this is because the elements
>> >> >>>>>> arrive at somewhat random order, and even if you do the
>> operation
>> >> >>>>>> "assign unique ID to elements" this might produce different
>> >> >>>>>> results when run multiple times)
>> >> >>>>> PCollections are *explicitly* unordered. Any operations that
>> >> >>>>> assume or
>> >> >>>>> depend on a specific ordering for correctness (or determinism)
>> must
>> >> >>>>> provide that ordering themselves (i.e. tolerate "arbitrary
>> shuffling
>> >> >>>>> of inputs"). As you point out, that may be very expensive if you
>> have
>> >> >>>>> very hot keys with very large (unbounded) timestamp skew.
>> >> >>>>>
>> >> >>>>> StatefulDoFns are low-level operations that should be used with
>> care;
>> >> >>>>> the simpler windowing model gives determinism in the face of
>> >> >>>>> unordered
>> >> >>>>> data (though late data and non-end-of-window triggering
>> introduces
>> >> >>>>> some of the non-determanism back in).
>> >> >>>>>
>> >> >>>>>> What worries me most is the property b), because it seems to me
>> >> >>>>>> to have serious consequences - not only that if you run twice
>> >> >>>>>> batch pipeline you would get different results, but even on
>> >> >>>>>> streaming, when pipeline fails and gets restarted from
>> >> >>>>>> checkpoint, produced output might differ from the previous run
>> >> >>>>>> and data from the first run might have already been persisted
>> >> >>>>>> into sink. That would create somewhat messy outputs.
>> >> >>>>> Beam has an exactly-once model. If the data was consumed, state
>> >> >>>>> mutated, and outputs written downstream (these three are
>> committed
>> >> >>>>> together atomically) it will not be replayed. That does not, of
>> >> >>>>> course, solve the non-determanism due to ordering (including the
>> fact
>> >> >>>>> that two operations reading the same PCollection may view
>> different
>> >> >>>>> ordering).
>> >> >>>>>
>> >> >>>>>> These two properties makes me think that the current
>> >> >>>>>> implementation is more of a _special case_ than the general one.
>> >> >>>>>> The general one would be that your state doesn't have the
>> >> >>>>>> properties to be able to tolerate buffering problems and/or
>> >> >>>>>> non-determinism. Which is the case where you need sorting in
>> both
>> >> >>>>>> streaming and batch to be part of the model.
>> >> >>>>>>
>> >> >>>>>> Let me point out one more analogy - that is merging vs.
>> >> >>>>>> non-merging windows. The general case (merging windows) implies
>> >> >>>>>> sorting by timestamp in both batch case (explicit) and streaming
>> >> >>>>>> (buffering). The special case (non-merging windows) doesn't rely
>> >> >>>>>> on any timestamp ordering, so the sorting and buffering can be
>> >> >>>>>> dropped. The underlying root cause of this is the same for both
>> >> >>>>>> stateful ParDo and windowing (essentially, assigning window
>> >> >>>>>> labels is a stateful operation when windowing function is
>> merging).
>> >> >>>>>>
>> >> >>>>>> The reason for the current behavior of stateful ParDo seems to
>> be
>> >> >>>>>> performance, but is it right to abandon correctness in favor of
>> >> >>>>>> performance? Wouldn't it be more consistent to have the default
>> >> >>>>>> behavior prefer correctness and when you have the specific
>> >> >>>>>> conditions of state function having special properties, then you
>> >> >>>>>> can annotate your DoFn (with something like
>> >> >>>>>> @TimeOrderingAgnostic), which would yield a better performance
>> in
>> >> >>>>>> that case?
>> >> >>>>> There are two separable questions here.
>> >> >>>>>
>> >> >>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are
>> received
>> >> >>>>> in a specific order? In the current model, it is not. Being able
>> to
>> >> >>>>> read, handle, and produced out-of-order data, including late
>> data, is
>> >> >>>>> a pretty fundamental property of distributed systems.
>> >> >>>>>
>> >> >>>>> 2) Given that some operations are easier (or possibly only
>> possible)
>> >> >>>>> to write when operating on ordered data, and that different
>> runners
>> >> >>>>> may have (significantly) cheaper ways to provide this ordering
>> than
>> >> >>>>> can be done by the user themselves, should we elevate this to a
>> >> >>>>> property of (Stateful?)DoFns that the runner can provide? I
>> think a
>> >> >>>>> compelling argument can be made here that we should.
>> >> >>>>>
>> >> >>>>> - Robert
>> >> >>>>>
>> >> >>>>>
>> >> >>>>>
>> >> >>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>> >> >>>>>>
>> >> >>>>>> Thanks for the nice small example of a calculation that depends
>> >> >>>>>> on order. You are right that many state machines have this
>> >> >>>>>> property. I agree w/ you and Luke that it is convenient for
>> batch
>> >> >>>>>> processing to sort by event timestamp before running a stateful
>> >> >>>>>> ParDo. In streaming you could also implement "sort by event
>> >> >>>>>> timestamp" by buffering until you know all earlier data will be
>> >> >>>>>> dropped - a slack buffer up to allowed lateness.
>> >> >>>>>>
>> >> >>>>>> I do not think that it is OK to sort in batch and not in
>> >> >>>>>> streaming. Many state machines diverge very rapidly when things
>> >> >>>>>> are out of order. So each runner if they see the
>> >> >>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
>> >> >>>>>> sorted data (by some mix of buffering and dropping), or to
>> reject
>> >> >>>>>> the pipeline as unsupported.
>> >> >>>>>>
>> >> >>>>>> And also want to say that this is not the default case - many
>> >> >>>>>> uses of state & timers in ParDo yield different results at the
>> >> >>>>>> element level, but the results are equivalent at in the big
>> >> >>>>>> picture. Such as the example of "assign a unique sequence number
>> >> >>>>>> to each element" or "group into batches" it doesn't matter
>> >> >>>>>> exactly what the result is, only that it meets the spec. And
>> >> >>>>>> other cases like user funnels are monotonic enough that you also
>> >> >>>>>> don't actually need sorting.
>> >> >>>>>>
>> >> >>>>>> Kenn
>> >> >>>>>>
>> >> >>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz>
>> >> >>>>>> wrote:
>> >> >>>>>>> Yes, the problem will arise probably mostly when you have not
>> >> >>>>>>> well distributed keys (or too few keys). I'm really not sure if
>> >> >>>>>>> a pure GBK with a trigger can solve this - it might help to
>> have
>> >> >>>>>>> data driven trigger. There would still be some doubts, though.
>> >> >>>>>>> The main question is still here - people say, that sorting by
>> >> >>>>>>> timestamp before stateful ParDo would be prohibitively slow,
>> but
>> >> >>>>>>> I don't really see why - the sorting is very probably already
>> >> >>>>>>> there. And if not (hash grouping instead of sorted grouping),
>> >> >>>>>>> then the sorting would affect only user defined StatefulParDos.
>> >> >>>>>>>
>> >> >>>>>>> This would suggest that the best way out of this would be
>> really
>> >> >>>>>>> to add annotation, so that the author of the pipeline can
>> decide.
>> >> >>>>>>>
>> >> >>>>>>> If that would be acceptable I think I can try to prepare some
>> >> >>>>>>> basic functionality, but I'm not sure, if I would be able to
>> >> >>>>>>> cover all runners / sdks.
>> >> >>>>>>>
>> >> >>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>> >> >>>>>>>
>> >> >>>>>>> It is read all per key and window and not just read all (this
>> >> >>>>>>> still won't scale with hot keys in the global window). The GBK
>> >> >>>>>>> preceding the StatefulParDo will guarantee that you are
>> >> >>>>>>> processing all the values for a specific key and window at any
>> >> >>>>>>> given time. Is there a specific window/trigger that is missing
>> >> >>>>>>> that you feel would remove the need for you to use
>> StatefulParDo?
>> >> >>>>>>>
>> >> >>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je.ik@seznam.cz
>> >
>> >> >>>>>>> wrote:
>> >> >>>>>>>> Hi Lukasz,
>> >> >>>>>>>>
>> >> >>>>>>>>> Today, if you must have a strict order, you must guarantee
>> >> >>>>>>>>> that your StatefulParDo implements the necessary "buffering &
>> >> >>>>>>>>> sorting" into state.
>> >> >>>>>>>> Yes, no problem with that. But this whole discussion started,
>> >> >>>>>>>> because *this doesn't work on batch*. You simply cannot first
>> >> >>>>>>>> read everything from distributed storage and then buffer it
>> all
>> >> >>>>>>>> into memory, just to read it again, but sorted. That will not
>> >> >>>>>>>> work. And even if it would, it would be a terrible waste of
>> >> >>>>>>>> resources.
>> >> >>>>>>>>
>> >> >>>>>>>> Jan
>> >> >>>>>>>>
>> >> >>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>> >> >>>>>>>>
>> >> >>>>>>>>
>> >> >>>>>>>>
>> >> >>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je.ik@seznam.cz
>> >
>> >> >>>>>>>> wrote:
>> >> >>>>>>>>> This discussion brings many really interesting questions for
>> >> >>>>>>>>> me. :-)
>> >> >>>>>>>>>
>> >> >>>>>>>>> > I don't see batch vs. streaming as part of the model.
>> One
>> >> >>>>>>>>> can have
>> >> >>>>>>>>> microbatch, or even a runner that alternates between
>> different
>> >> >>>>>>>>> modes.
>> >> >>>>>>>>>
>> >> >>>>>>>>> Although I understand motivation of this statement, this
>> >> >>>>>>>>> project name is
>> >> >>>>>>>>> "Apache Beam: An advanced unified programming model". What
>> >> >>>>>>>>> does the
>> >> >>>>>>>>> model unify, if "streaming vs. batch" is not part of the
>> model?
>> >> >>>>>>>>>
>> >> >>>>>>>>> Using microbatching, chaining of batch jobs, or pure
>> streaming
>> >> >>>>>>>>> are
>> >> >>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
>> >> >>>>>>>>> All these
>> >> >>>>>>>>> define several runtime parameters, which in turn define how
>> >> >>>>>>>>> well/badly
>> >> >>>>>>>>> will the pipeline perform and how many resources might be
>> >> >>>>>>>>> needed. From
>> >> >>>>>>>>> my point of view, pure streaming should be the most resource
>> >> >>>>>>>>> demanding
>> >> >>>>>>>>> (if not, why bother with batch? why not run everything in
>> >> >>>>>>>>> streaming
>> >> >>>>>>>>> only? what will there remain to "unify"?).
>> >> >>>>>>>>>
>> >> >>>>>>>>> > Fortunately, for batch, only the state for a single key
>> >> >>>>>>>>> needs to be
>> >> >>>>>>>>> preserved at a time, rather than the state for all keys
>> across
>> >> >>>>>>>>> the range
>> >> >>>>>>>>> of skew. Of course if you have few or hot keys, one can still
>> >> >>>>>>>>> have
>> >> >>>>>>>>> issues (and this is not specific to StatefulDoFns).
>> >> >>>>>>>>>
>> >> >>>>>>>>> Yes, but here is still the presumption that my stateful DoFn
>> can
>> >> >>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the
>> use
>> >> >>>>>>>>> case in
>> >> >>>>>>>>> more detail.
>> >> >>>>>>>>>
>> >> >>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
>> >> >>>>>>>>> some key for
>> >> >>>>>>>>> each element, which is irrelevant for the demonstration).
>> Your
>> >> >>>>>>>>> task is
>> >> >>>>>>>>> to calculate in running global window the actual number of
>> >> >>>>>>>>> changes
>> >> >>>>>>>>> between state 0 and state 1 and vice versa. When the state
>> >> >>>>>>>>> doesn't
>> >> >>>>>>>>> change, you don't calculate anything. If input (for given
>> key)
>> >> >>>>>>>>> would be
>> >> >>>>>>>>> (tN denotes timestamp N):
>> >> >>>>>>>>>
>> >> >>>>>>>>> t1: 1
>> >> >>>>>>>>>
>> >> >>>>>>>>> t2: 0
>> >> >>>>>>>>>
>> >> >>>>>>>>> t3: 0
>> >> >>>>>>>>>
>> >> >>>>>>>>> t4: 1
>> >> >>>>>>>>>
>> >> >>>>>>>>> t5: 1
>> >> >>>>>>>>>
>> >> >>>>>>>>> t6: 0
>> >> >>>>>>>>>
>> >> >>>>>>>>> then the output should yield (supposing that default state is
>> >> >>>>>>>>> zero):
>> >> >>>>>>>>>
>> >> >>>>>>>>> t1: (one: 1, zero: 0)
>> >> >>>>>>>>>
>> >> >>>>>>>>> t2: (one: 1, zero: 1)
>> >> >>>>>>>>>
>> >> >>>>>>>>> t3: (one: 1, zero: 1)
>> >> >>>>>>>>>
>> >> >>>>>>>>> t4: (one: 2, zero: 1)
>> >> >>>>>>>>>
>> >> >>>>>>>>> t5: (one: 2, zero: 1)
>> >> >>>>>>>>>
>> >> >>>>>>>>> t6: (one: 2, zero: 2)
>> >> >>>>>>>>>
>> >> >>>>>>>>> How would you implement this in current Beam semantics?
>> >> >>>>>>>> I think your saying here that I know that my input is ordered
>> >> >>>>>>>> in a specific way and since I assume the order when writing my
>> >> >>>>>>>> pipeline I can perform this optimization. But there is nothing
>> >> >>>>>>>> preventing a runner from noticing that your processing in the
>> >> >>>>>>>> global window with a specific type of trigger and re-ordering
>> >> >>>>>>>> your inputs/processing to get better performance (since you
>> >> >>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>> >> >>>>>>>> streaming for the GlobalWindow).
>> >> >>>>>>>>
>> >> >>>>>>>> Today, if you must have a strict order, you must guarantee
>> that
>> >> >>>>>>>> your StatefulParDo implements the necessary "buffering &
>> >> >>>>>>>> sorting" into state. I can see why you would want an
>> annotation
>> >> >>>>>>>> that says I must have timestamp ordered elements, since it
>> >> >>>>>>>> makes writing certain StatefulParDos much easier.
>> StatefulParDo
>> >> >>>>>>>> is a low-level function, it really is the "here you go and do
>> >> >>>>>>>> whatever you need to but here be dragons" function while
>> >> >>>>>>>> windowing and triggering is meant to keep many people from
>> >> >>>>>>>> writing StatefulParDo in the first place.
>> >> >>>>>>>>
>> >> >>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
>> >> >>>>>>>>> are likely to
>> >> >>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>> >> >>>>>>>>> falls
>> >> >>>>>>>>> behind in streaming mode as well.
>> >> >>>>>>>>>
>> >> >>>>>>>>> But the worst case is defined by input of size (available
>> >> >>>>>>>>> resources +
>> >> >>>>>>>>> single byte) -> pipeline fail. Although it could have
>> >> >>>>>>>>> finished, given
>> >> >>>>>>>>> the right conditions.
>> >> >>>>>>>>>
>> >> >>>>>>>>> > This might be reasonable, implemented by default by
>> >> >>>>>>>>> buffering
>> >> >>>>>>>>> everything and releasing elements as the watermark
>> (+lateness)
>> >> >>>>>>>>> advances,
>> >> >>>>>>>>> but would likely lead to inefficient (though *maybe* easier
>> to
>> >> >>>>>>>>> reason
>> >> >>>>>>>>> about) code.
>> >> >>>>>>>>>
>> >> >>>>>>>>> Sure, the pipeline will be less efficient, because it would
>> >> >>>>>>>>> have to
>> >> >>>>>>>>> buffer and sort the inputs. But at least it will produce
>> >> >>>>>>>>> correct results
>> >> >>>>>>>>> in cases where updates to state are order-sensitive.
>> >> >>>>>>>>>
>> >> >>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
>> >> >>>>>>>>> (key, values):
>> >> >>>>>>>>> [(key, value) for value in values])?
>> >> >>>>>>>>>
>> >> >>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
>> >> >>>>>>>>> The trigger
>> >> >>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
>> >> >>>>>>>>> element
>> >> >>>>>>>>> with lowest timestamp in the buffer. Although this could be
>> >> >>>>>>>>> somehow
>> >> >>>>>>>>> emulated by fixed trigger each X millis.
>> >> >>>>>>>>>
>> >> >>>>>>>>> > Or is the underlying desire just to be able to hint to
>> >> >>>>>>>>> the runner
>> >> >>>>>>>>> that the code may perform better (e.g. require less
>> resources)
>> >> >>>>>>>>> as skew
>> >> >>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
>> >> >>>>>>>>>
>> >> >>>>>>>>> No, the sorting would have to be done in streaming case as
>> >> >>>>>>>>> well. That is
>> >> >>>>>>>>> an imperative of the unified model. I think it is possible to
>> >> >>>>>>>>> sort by
>> >> >>>>>>>>> timestamp only in batch case (and do it for *all* batch
>> >> >>>>>>>>> stateful pardos
>> >> >>>>>>>>> without annotation), or introduce annotation, but then make
>> >> >>>>>>>>> the same
>> >> >>>>>>>>> guarantees for streaming case as well.
>> >> >>>>>>>>>
>> >> >>>>>>>>> Jan
>> >> >>>>>>>>>
>> >> >>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>> >> >>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>> >> >>>>>>>>>> <je...@seznam.cz> wrote:
>> >> >>>>>>>>>>> Hi Robert,
>> >> >>>>>>>>>>>
>> >> >>>>>>>>>>> yes, I think you rephrased my point - although no
>> *explicit*
>> >> >>>>>>>>>>> guarantees
>> >> >>>>>>>>>>> of ordering are given in either mode, there is *implicit*
>> >> >>>>>>>>>>> ordering in
>> >> >>>>>>>>>>> streaming case that is due to nature of the processing -
>> the
>> >> >>>>>>>>>>> difference
>> >> >>>>>>>>>>> between watermark and timestamp of elements flowing through
>> >> >>>>>>>>>>> the pipeline
>> >> >>>>>>>>>>> are generally low (too high difference leads to the
>> >> >>>>>>>>>>> overbuffering
>> >> >>>>>>>>>>> problem), but there is no such bound on batch.
>> >> >>>>>>>>>> Fortunately, for batch, only the state for a single key
>> needs
>> >> >>>>>>>>>> to be
>> >> >>>>>>>>>> preserved at a time, rather than the state for all keys
>> >> >>>>>>>>>> across the
>> >> >>>>>>>>>> range of skew. Of course if you have few or hot keys, one
>> can
>> >> >>>>>>>>>> still
>> >> >>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>> >> >>>>>>>>>>
>> >> >>>>>>>>>>> As a result, I see a few possible solutions:
>> >> >>>>>>>>>>>
>> >> >>>>>>>>>>> - the best and most natural seems to be extension of
>> >> >>>>>>>>>>> the model, so
>> >> >>>>>>>>>>> that it defines batch as not only "streaming pipeline
>> >> >>>>>>>>>>> executed in batch
>> >> >>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>> >> >>>>>>>>>>> characteristics as
>> >> >>>>>>>>>>> in streaming case, executed in batch fashion", I really
>> >> >>>>>>>>>>> don't think that
>> >> >>>>>>>>>>> there are any conflicts with the current model, or that
>> this
>> >> >>>>>>>>>>> could
>> >> >>>>>>>>>>> affect performance, because the required sorting (as
>> pointed by
>> >> >>>>>>>>>>> Aljoscha) is very probably already done during translation
>> >> >>>>>>>>>>> of stateful
>> >> >>>>>>>>>>> pardos. Also note that this definition only affects user
>> >> >>>>>>>>>>> defined
>> >> >>>>>>>>>>> stateful pardos
>> >> >>>>>>>>>> I don't see batch vs. streaming as part of the model. One
>> can
>> >> >>>>>>>>>> have
>> >> >>>>>>>>>> microbatch, or even a runner that alternates between
>> >> >>>>>>>>>> different modes.
>> >> >>>>>>>>>> The model describes what the valid outputs are given a
>> >> >>>>>>>>>> (sometimes
>> >> >>>>>>>>>> partial) set of inputs. It becomes really hard to define
>> >> >>>>>>>>>> things like
>> >> >>>>>>>>>> "as good runtime characteristics." Once you allow any
>> >> >>>>>>>>>> out-of-orderedness, it is not very feasible to try and
>> define
>> >> >>>>>>>>>> (and
>> >> >>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>> >> >>>>>>>>>> out-of-orderedness.
>> >> >>>>>>>>>>
>> >> >>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
>> >> >>>>>>>>>> likely to
>> >> >>>>>>>>>> degrade poorly (possibly catastrophically) when the
>> watermark
>> >> >>>>>>>>>> falls
>> >> >>>>>>>>>> behind in streaming mode as well.
>> >> >>>>>>>>>>
>> >> >>>>>>>>>>> - another option would be to introduce annotation for
>> >> >>>>>>>>>>> DoFns (e.g.
>> >> >>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
>> >> >>>>>>>>>>> the sorting
>> >> >>>>>>>>>>> in batch case - but - this extension would have to ensure
>> >> >>>>>>>>>>> the sorting in
>> >> >>>>>>>>>>> streaming mode also - it would require definition of
>> allowed
>> >> >>>>>>>>>>> lateness,
>> >> >>>>>>>>>>> and triggger (essentially similar to window)
>> >> >>>>>>>>>> This might be reasonable, implemented by default by
>> buffering
>> >> >>>>>>>>>> everything and releasing elements as the watermark
>> (+lateness)
>> >> >>>>>>>>>> advances, but would likely lead to inefficient (though
>> >> >>>>>>>>>> *maybe* easier
>> >> >>>>>>>>>> to reason about) code. Not sure about the semantics of
>> >> >>>>>>>>>> triggering
>> >> >>>>>>>>>> here, especially data-driven triggers. Would it be roughly
>> >> >>>>>>>>>> equivalent
>> >> >>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
>> >> >>>>>>>>>> value in
>> >> >>>>>>>>>> values])?
>> >> >>>>>>>>>>
>> >> >>>>>>>>>> Or is the underlying desire just to be able to hint to the
>> >> >>>>>>>>>> runner that
>> >> >>>>>>>>>> the code may perform better (e.g. require less resources) as
>> >> >>>>>>>>>> skew is
>> >> >>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>> >> >>>>>>>>>>
>> >> >>>>>>>>>>> - last option would be to introduce these "higher
>> order
>> >> >>>>>>>>>>> guarantees" in
>> >> >>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be
>> the
>> >> >>>>>>>>>>> worst
>> >> >>>>>>>>>>> option to me
>> >> >>>>>>>>>>>
>> >> >>>>>>>>>>> I see the first two options quite equally good, although
>> the
>> >> >>>>>>>>>>> letter one
>> >> >>>>>>>>>>> is probably more time consuming to implement. But it would
>> >> >>>>>>>>>>> bring
>> >> >>>>>>>>>>> additional feature to streaming case as well.
>> >> >>>>>>>>>>>
>> >> >>>>>>>>>>> Thanks for any thoughts.
>> >> >>>>>>>>>>>
>> >> >>>>>>>>>>> Jan
>> >> >>>>>>>>>>>
>> >> >>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>> >> >>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>> >> >>>>>>>>>>>> <je...@seznam.cz> wrote:
>> >> >>>>>>>>>>>>> Hi Reuven,
>> >> >>>>>>>>>>>>>
>> >> >>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
>> >> >>>>>>>>>>>>>> runners.
>> >> >>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
>> >> >>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
>> >> >>>>>>>>>>>>> of elements. That basically (practically) can work only
>> >> >>>>>>>>>>>>> for cases, where the order of input elements doesn't
>> >> >>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and
>> any
>> >> >>>>>>>>>>>>> time you have a state machine involved, then the ordering
>> >> >>>>>>>>>>>>> of elements would matter.
>> >> >>>>>>>>>>>> No guarantees on order are provided in *either* streaming
>> >> >>>>>>>>>>>> or batch
>> >> >>>>>>>>>>>> mode by the model. However, it is the case that in order
>> to
>> >> >>>>>>>>>>>> make
>> >> >>>>>>>>>>>> forward progress most streaming runners attempt to limit
>> >> >>>>>>>>>>>> the amount of
>> >> >>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
>> >> >>>>>>>>>>>> processing
>> >> >>>>>>>>>>>> time) to make forward progress, which in turn could help
>> >> >>>>>>>>>>>> cap the
>> >> >>>>>>>>>>>> amount of state that must be held concurrently, whereas a
>> >> >>>>>>>>>>>> batch runner
>> >> >>>>>>>>>>>> may not allow any state to be safely discarded until the
>> whole
>> >> >>>>>>>>>>>> timeline from infinite past to infinite future has been
>> >> >>>>>>>>>>>> observed.
>> >> >>>>>>>>>>>>
>> >> >>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
>> >> >>>>>>>>>>>> batch" in batch mode.
>> >> >>>>>>>>>>>>
>> >> >>>>>>>>>>>>
>> >> >>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>> >> >>>>>>>>>>>> <mx...@apache.org> wrote:
>> >> >>>>>>>>>>>>
>> >> >>>>>>>>>>>>>> batch semantics and streaming semantics differs
>> only
>> >> >>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
>> >> >>>>>>>>>>>>>> batch and cannot on stream
>> >> >>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
>> >> >>>>>>>>>>>>> trigger. You
>> >> >>>>>>>>>>>>> could define additional triggers that do early firings.
>> >> >>>>>>>>>>>>> And you could
>> >> >>>>>>>>>>>>> even trigger the global window by advancing the watermark
>> >> >>>>>>>>>>>>> to +inf.
>> >> >>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window
>> with
>> >> >>>>>>>>>>>> default
>> >> >>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
>> >> >>>>>>>>>>>> is more
>> >> >>>>>>>>>>>> likely to be user error than an actual desire to have no
>> >> >>>>>>>>>>>> output until
>> >> >>>>>>>>>>>> drain. But it's semantically valid in the model.
>>
>>
Re: Definition of Unified model
Posted by Reuven Lax <re...@google.com>.
Not really. I'm suggesting that some variant of FIFO ordering is necessary,
which requires either runners natively support FIFO ordering or transforms
adding some extra sequence number to each record to sort by.
I still think your proposal is very useful by the way. I'm merely pointing
out that to solve the state-machine problem we probably need something more.
Reuven
On Thu, May 23, 2019 at 9:50 AM Jan Lukavský <je...@seznam.cz> wrote:
> Hi,
> yes. It seems that ordering by user supplied UDF makes sense and I will
> update the design proposal accordingly.
> Would that solve the issues you mention?
> Jan
> ---------- Původní e-mail ----------
> Od: Reuven Lax <re...@google.com>
> Komu: dev <de...@beam.apache.org>
> Datum: 23. 5. 2019 18:44:38
> Předmět: Re: Definition of Unified model
>
> I'm simply saying that timestamp ordering is insufficient for state
> machines. I wasn't proposing Kafka as a solution - that was simply an
> example of how people solve this problem in other scenarios.
>
> BTW another example of ordering: Imagine today that you have a triggered
> Sum aggregation writing out to a key-value sink. In theory we provide no
> ordering, so the sink might write the triggered sums in the wrong order,
> ending up with an incorrect value in the sink. In this case you probably
> want values ordered by trigger pane index.
>
> Reuven
>
> On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz> wrote:
>
> Hi Reuven,
> I share the view point of Robert. I think the isuue you refer to is not in
> reality related to timestamps, but to the fact, that ordering of events in
> time is observer dependent (either caused by relativity, or time skew,
> essentially this has the same consequences). And the resolution in fact
> isn't Kafka, but generally an authoritative observer, that tells you "I saw
> the events in this order". And you either have one (and have the outcome of
> his observation persisted in the data - e.g. as offset in Kafka partition),
> then you should be able to use it (maybe that suggests afterall that
> sorting by some user supplied UDF might make sense), or do not have it, and
> then any interpretation of the data seems to be equally valid. Although
> determinism is fine, of course.
> Jan
> ---------- Původní e-mail ----------
> Od: Reuven Lax <re...@google.com>
> Komu: dev <de...@beam.apache.org>
> Datum: 23. 5. 2019 17:39:12
> Předmět: Re: Definition of Unified model
>
> So an example would be elements of type "startUserSession" and
> "endUserSession" (website sessions, not Beam sessions). Logically you may
> need to process them in the correct order if you have any sort of
> state-machine logic. However timestamp ordering is never guaranteed to
> match the logical ordering. Not only might you have several elements with
> the same timestamp, but in reality time skew across backend servers can
> cause the events to have timestamps in reverse order of the actual
> causality order.
>
> People do solve this problem today though. Publish the events to Kafka,
> making sure that events for the same user end up in the same Kafka
> partition. This ensures that the events appear in the Kafka partitions in
> causality order, even if the timestamp order doesn't match. The your Kafka
> subscriber simply process the elements in each partition in order.
>
> I think the ability to impose FIFO causality ordering is what's needed for
> any state-machine work. Timestamp ordering has advantages (though often I
> think the advantage is in state), but does not solve this problem.
>
> Reuven
>
> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <ro...@google.com>
> wrote:
>
> Good point.
>
> The "implementation-specific" way I would do this is
> window-by-instant, followed by a DoFn that gets all the elements with
> the same timestamp and sorts/acts accordingly, but this counts on the
> runner producing windows in timestamp order (likely?) and also the
> subsequent DoFn getting them in this order (also likely, due to
> fusion).
>
> One could make the argument that, though it does not provide
> deterministic behavior, getting elements of the same timestamp in
> different orders should produce equally valid interpretations of the
> data. (After all, due to relatively, timestamps are not technically
> well ordered across space.) I can see how data-dependent tiebreakers
> could be useful, or promises of preservation of order between
> operations.
>
> - Robert
>
> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com> wrote:
> >
> > So Jan's example of state machines is quite a valid use case for
> ordering. However in my experience, timestamp ordering is insufficient for
> state machines. Elements that cause state transitions might come in with
> the exact same timestamp, yet still have a necessary ordering. Especially
> given Beam's decision to have milliseconds timestamps this is possible, but
> even at microsecond or nanosecond precision this can happen at scale. To
> handle state machines you usually need some sort of FIFO ordering along
> with an ordered sources, such as Kafka, not timestamp ordering.
> >
> > Reuven
> >
> > On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz> wrote:
> >>
> >> Hi all,
> >>
> >> thanks everyone for this discussion. I think I have gathered enough
> >> feedback to be able to put down a proposition for changes, which I will
> >> do and send to this list for further discussion. There are still doubts
> >> remaining the non-determinism and it's relation to outputs stability vs.
> >> latency. But I will try to clarify all this in the design document.
> >>
> >> Thanks,
> >>
> >> Jan
> >>
> >> On 5/22/19 3:49 PM, Maximilian Michels wrote:
> >> >> Someone from Flink might correct me if I'm wrong, but that's my
> >> >> current understanding.
> >> >
> >> > In essence your description of how exactly-once works in Flink is
> >> > correct. The general assumption in Flink is that pipelines must be
> >> > deterministic and thus produce idempotent writes in the case of
> >> > failures. However, that doesn't mean Beam sinks can't guarantee a bit
> >> > more with what Flink has to offer.
> >> >
> >> > Luke already mentioned the design discussions for @RequiresStableInput
> >> > which ensures idempotent writes for non-deterministic pipelines. This
> >> > is not part of the model but an optional Beam feature.
> >> >
> >> > We recently implemented support for @RequiresStableInput in the Flink
> >> > Runner. Reuven mentioned the Flink checkpoint confirmation, which
> >> > allows us to buffer (and checkpoint) processed data and only emit it
> >> > once a Flink checkpoint has completed.
> >> >
> >> > Cheers,
> >> > Max
> >> >
> >> > On 21.05.19 16:49, Jan Lukavský wrote:
> >> >> Hi,
> >> >>
> >> >> > Actually, I think it is a larger (open) question whether exactly
> >> >> once is guaranteed by the model or whether runners are allowed to
> >> >> relax that. I would think, however, that sources correctly
> >> >> implemented should be idempotent when run atop an exactly once
> >> >> infrastructure such as Flink of Dataflow.
> >> >>
> >> >> I would assume, that the model basically inherits guarantees of
> >> >> underlying infrastructure. Because Flink does not work as you
> >> >> described (atomic commit of inputs, state and outputs), but rather a
> >> >> checkpoint mark is flowing through the DAG much like watermark and on
> >> >> failures operators are restored and data reprocessed, it (IMHO)
> >> >> implies, that you have exactly once everywhere in the DAG *but*
> >> >> sinks. That is because sinks cannot be restored to previous state,
> >> >> instead sinks are supposed to be idempotent in order for the exactly
> >> >> once to really work (or at least be able to commit outputs on
> >> >> checkpoint in sink). That implies that if you don't have sink that is
> >> >> able to commit outputs atomically on checkpoint, the pipeline
> >> >> execution should be deterministic upon retries, otherwise shadow
> >> >> writes from failed paths of the pipeline might appear.
> >> >>
> >> >> Someone from Flink might correct me if I'm wrong, but that's my
> >> >> current understanding.
> >> >>
> >> >> > Sounds like we should make this clearer.
> >> >>
> >> >> I meant that you are right that we must not in any thoughts we are
> >> >> having forget that streams are by definition out-of-order. That is
> >> >> property that we cannot change. But - that doesn't limit us from
> >> >> creating operator that presents the data to UDF as if the stream was
> >> >> ideally sorted. It can do that by introducing latency, of course.
> >> >>
> >> >> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
> >> >>> Reza: One could provide something like this as a utility class, but
> >> >>> one downside is that it is not scale invariant. It requires a tuning
> >> >>> parameter that, if to small, won't mitigate the problem, but if to
> >> >>> big, greatly increases latency. (Possibly one could define a dynamic
> >> >>> session-like window to solve this though...) It also might be harder
> >> >>> for runners that *can* cheaply present stuff in timestamp order to
> >> >>> optimize. (That and, in practice, our annotation-style process
> methods
> >> >>> don't lend themselves to easy composition.) I think it could work in
> >> >>> specific cases though.
> >> >>>
> >> >>> More inline below.
> >> >>>
> >> >>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz>
> wrote:
> >> >>>> Hi Robert,
> >> >>>>
> >> >>>> > Beam has an exactly-once model. If the data was consumed, state
> >> >>>> mutated, and outputs written downstream (these three are committed
> >> >>>> together atomically) it will not be replayed. That does not, of
> >> >>>> course,
> >> >>>> solve the non-determanism due to ordering (including the fact that
> two
> >> >>>> operations reading the same PCollection may view different
> ordering).
> >> >>>>
> >> >>>> I think what you describe is a property of a runner, not of the
> model,
> >> >>>> right? I think if I run my pipeline on Flink I will not get this
> >> >>>> atomicity, because although Flink uses also exactly-once model if
> >> >>>> might
> >> >>>> write outputs multiple times.
> >> >>> Actually, I think it is a larger (open) question whether exactly
> once
> >> >>> is guaranteed by the model or whether runners are allowed to relax
> >> >>> that. I would think, however, that sources correctly implemented
> >> >>> should be idempotent when run atop an exactly once infrastructure
> such
> >> >>> as Flink of Dataflow.
> >> >>>
> >> >>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
> >> >>>> received
> >> >>>> in a specific order? In the current model, it is not. Being able to
> >> >>>> read, handle, and produced out-of-order data, including late data,
> >> >>>> is a
> >> >>>> pretty fundamental property of distributed systems.
> >> >>>>
> >> >>>> Yes, absolutely. The argument here is not that Stateful ParDo
> should
> >> >>>> presume to receive elements in any order, but to _present_ it as
> >> >>>> such to
> >> >>>> the user @ProcessElement function.
> >> >>> Sounds like we should make this clearer.
> >> >>>
> >> >>>> > 2) Given that some operations are easier (or possibly only
> >> >>>> possible)
> >> >>>> to write when operating on ordered data, and that different runners
> >> >>>> may
> >> >>>> have (significantly) cheaper ways to provide this ordering than
> can be
> >> >>>> done by the user themselves, should we elevate this to a property
> of
> >> >>>> (Stateful?)DoFns that the runner can provide? I think a compelling
> >> >>>> argument can be made here that we should.
> >> >>>>
> >> >>>> +1
> >> >>>>
> >> >>>> Jan
> >> >>>>
> >> >>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
> >> >>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >> >>>>>> > I don't see batch vs. streaming as part of the model. One
> >> >>>>>> can have
> >> >>>>>> microbatch, or even a runner that alternates between different
> >> >>>>>> modes.
> >> >>>>>>
> >> >>>>>> Although I understand motivation of this statement, this project
> >> >>>>>> name is
> >> >>>>>> "Apache Beam: An advanced unified programming model". What does
> the
> >> >>>>>> model unify, if "streaming vs. batch" is not part of the model?
> >> >>>>> What I mean is that streaming vs. batch is no longer part of the
> >> >>>>> model
> >> >>>>> (or ideally API), but pushed down to be a concern of the runner
> >> >>>>> (executor) of the pipeline.
> >> >>>>>
> >> >>>>>
> >> >>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz>
> >> >>>>> wrote:
> >> >>>>>> Hi Kenn,
> >> >>>>>>
> >> >>>>>> OK, so if we introduce annotation, we can have stateful ParDo
> >> >>>>>> with sorting, that would perfectly resolve my issues. I still
> >> >>>>>> have some doubts, though. Let me explain. The current behavior of
> >> >>>>>> stateful ParDo has the following properties:
> >> >>>>>>
> >> >>>>>> a) might fail in batch, although runs fine in streaming (that
> >> >>>>>> is due to the buffering, and unbounded lateness in batch, which
> >> >>>>>> was discussed back and forth in this thread)
> >> >>>>>>
> >> >>>>>> b) might be non deterministic (this is because the elements
> >> >>>>>> arrive at somewhat random order, and even if you do the operation
> >> >>>>>> "assign unique ID to elements" this might produce different
> >> >>>>>> results when run multiple times)
> >> >>>>> PCollections are *explicitly* unordered. Any operations that
> >> >>>>> assume or
> >> >>>>> depend on a specific ordering for correctness (or determinism)
> must
> >> >>>>> provide that ordering themselves (i.e. tolerate "arbitrary
> shuffling
> >> >>>>> of inputs"). As you point out, that may be very expensive if you
> have
> >> >>>>> very hot keys with very large (unbounded) timestamp skew.
> >> >>>>>
> >> >>>>> StatefulDoFns are low-level operations that should be used with
> care;
> >> >>>>> the simpler windowing model gives determinism in the face of
> >> >>>>> unordered
> >> >>>>> data (though late data and non-end-of-window triggering introduces
> >> >>>>> some of the non-determanism back in).
> >> >>>>>
> >> >>>>>> What worries me most is the property b), because it seems to me
> >> >>>>>> to have serious consequences - not only that if you run twice
> >> >>>>>> batch pipeline you would get different results, but even on
> >> >>>>>> streaming, when pipeline fails and gets restarted from
> >> >>>>>> checkpoint, produced output might differ from the previous run
> >> >>>>>> and data from the first run might have already been persisted
> >> >>>>>> into sink. That would create somewhat messy outputs.
> >> >>>>> Beam has an exactly-once model. If the data was consumed, state
> >> >>>>> mutated, and outputs written downstream (these three are committed
> >> >>>>> together atomically) it will not be replayed. That does not, of
> >> >>>>> course, solve the non-determanism due to ordering (including the
> fact
> >> >>>>> that two operations reading the same PCollection may view
> different
> >> >>>>> ordering).
> >> >>>>>
> >> >>>>>> These two properties makes me think that the current
> >> >>>>>> implementation is more of a _special case_ than the general one.
> >> >>>>>> The general one would be that your state doesn't have the
> >> >>>>>> properties to be able to tolerate buffering problems and/or
> >> >>>>>> non-determinism. Which is the case where you need sorting in both
> >> >>>>>> streaming and batch to be part of the model.
> >> >>>>>>
> >> >>>>>> Let me point out one more analogy - that is merging vs.
> >> >>>>>> non-merging windows. The general case (merging windows) implies
> >> >>>>>> sorting by timestamp in both batch case (explicit) and streaming
> >> >>>>>> (buffering). The special case (non-merging windows) doesn't rely
> >> >>>>>> on any timestamp ordering, so the sorting and buffering can be
> >> >>>>>> dropped. The underlying root cause of this is the same for both
> >> >>>>>> stateful ParDo and windowing (essentially, assigning window
> >> >>>>>> labels is a stateful operation when windowing function is
> merging).
> >> >>>>>>
> >> >>>>>> The reason for the current behavior of stateful ParDo seems to be
> >> >>>>>> performance, but is it right to abandon correctness in favor of
> >> >>>>>> performance? Wouldn't it be more consistent to have the default
> >> >>>>>> behavior prefer correctness and when you have the specific
> >> >>>>>> conditions of state function having special properties, then you
> >> >>>>>> can annotate your DoFn (with something like
> >> >>>>>> @TimeOrderingAgnostic), which would yield a better performance in
> >> >>>>>> that case?
> >> >>>>> There are two separable questions here.
> >> >>>>>
> >> >>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are
> received
> >> >>>>> in a specific order? In the current model, it is not. Being able
> to
> >> >>>>> read, handle, and produced out-of-order data, including late
> data, is
> >> >>>>> a pretty fundamental property of distributed systems.
> >> >>>>>
> >> >>>>> 2) Given that some operations are easier (or possibly only
> possible)
> >> >>>>> to write when operating on ordered data, and that different
> runners
> >> >>>>> may have (significantly) cheaper ways to provide this ordering
> than
> >> >>>>> can be done by the user themselves, should we elevate this to a
> >> >>>>> property of (Stateful?)DoFns that the runner can provide? I think
> a
> >> >>>>> compelling argument can be made here that we should.
> >> >>>>>
> >> >>>>> - Robert
> >> >>>>>
> >> >>>>>
> >> >>>>>
> >> >>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
> >> >>>>>>
> >> >>>>>> Thanks for the nice small example of a calculation that depends
> >> >>>>>> on order. You are right that many state machines have this
> >> >>>>>> property. I agree w/ you and Luke that it is convenient for batch
> >> >>>>>> processing to sort by event timestamp before running a stateful
> >> >>>>>> ParDo. In streaming you could also implement "sort by event
> >> >>>>>> timestamp" by buffering until you know all earlier data will be
> >> >>>>>> dropped - a slack buffer up to allowed lateness.
> >> >>>>>>
> >> >>>>>> I do not think that it is OK to sort in batch and not in
> >> >>>>>> streaming. Many state machines diverge very rapidly when things
> >> >>>>>> are out of order. So each runner if they see the
> >> >>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
> >> >>>>>> sorted data (by some mix of buffering and dropping), or to reject
> >> >>>>>> the pipeline as unsupported.
> >> >>>>>>
> >> >>>>>> And also want to say that this is not the default case - many
> >> >>>>>> uses of state & timers in ParDo yield different results at the
> >> >>>>>> element level, but the results are equivalent at in the big
> >> >>>>>> picture. Such as the example of "assign a unique sequence number
> >> >>>>>> to each element" or "group into batches" it doesn't matter
> >> >>>>>> exactly what the result is, only that it meets the spec. And
> >> >>>>>> other cases like user funnels are monotonic enough that you also
> >> >>>>>> don't actually need sorting.
> >> >>>>>>
> >> >>>>>> Kenn
> >> >>>>>>
> >> >>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz>
> >> >>>>>> wrote:
> >> >>>>>>> Yes, the problem will arise probably mostly when you have not
> >> >>>>>>> well distributed keys (or too few keys). I'm really not sure if
> >> >>>>>>> a pure GBK with a trigger can solve this - it might help to have
> >> >>>>>>> data driven trigger. There would still be some doubts, though.
> >> >>>>>>> The main question is still here - people say, that sorting by
> >> >>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
> >> >>>>>>> I don't really see why - the sorting is very probably already
> >> >>>>>>> there. And if not (hash grouping instead of sorted grouping),
> >> >>>>>>> then the sorting would affect only user defined StatefulParDos.
> >> >>>>>>>
> >> >>>>>>> This would suggest that the best way out of this would be really
> >> >>>>>>> to add annotation, so that the author of the pipeline can
> decide.
> >> >>>>>>>
> >> >>>>>>> If that would be acceptable I think I can try to prepare some
> >> >>>>>>> basic functionality, but I'm not sure, if I would be able to
> >> >>>>>>> cover all runners / sdks.
> >> >>>>>>>
> >> >>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
> >> >>>>>>>
> >> >>>>>>> It is read all per key and window and not just read all (this
> >> >>>>>>> still won't scale with hot keys in the global window). The GBK
> >> >>>>>>> preceding the StatefulParDo will guarantee that you are
> >> >>>>>>> processing all the values for a specific key and window at any
> >> >>>>>>> given time. Is there a specific window/trigger that is missing
> >> >>>>>>> that you feel would remove the need for you to use
> StatefulParDo?
> >> >>>>>>>
> >> >>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
> >> >>>>>>> wrote:
> >> >>>>>>>> Hi Lukasz,
> >> >>>>>>>>
> >> >>>>>>>>> Today, if you must have a strict order, you must guarantee
> >> >>>>>>>>> that your StatefulParDo implements the necessary "buffering &
> >> >>>>>>>>> sorting" into state.
> >> >>>>>>>> Yes, no problem with that. But this whole discussion started,
> >> >>>>>>>> because *this doesn't work on batch*. You simply cannot first
> >> >>>>>>>> read everything from distributed storage and then buffer it all
> >> >>>>>>>> into memory, just to read it again, but sorted. That will not
> >> >>>>>>>> work. And even if it would, it would be a terrible waste of
> >> >>>>>>>> resources.
> >> >>>>>>>>
> >> >>>>>>>> Jan
> >> >>>>>>>>
> >> >>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
> >> >>>>>>>>
> >> >>>>>>>>
> >> >>>>>>>>
> >> >>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
> >> >>>>>>>> wrote:
> >> >>>>>>>>> This discussion brings many really interesting questions for
> >> >>>>>>>>> me. :-)
> >> >>>>>>>>>
> >> >>>>>>>>> > I don't see batch vs. streaming as part of the model. One
> >> >>>>>>>>> can have
> >> >>>>>>>>> microbatch, or even a runner that alternates between different
> >> >>>>>>>>> modes.
> >> >>>>>>>>>
> >> >>>>>>>>> Although I understand motivation of this statement, this
> >> >>>>>>>>> project name is
> >> >>>>>>>>> "Apache Beam: An advanced unified programming model". What
> >> >>>>>>>>> does the
> >> >>>>>>>>> model unify, if "streaming vs. batch" is not part of the
> model?
> >> >>>>>>>>>
> >> >>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
> >> >>>>>>>>> are
> >> >>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
> >> >>>>>>>>> All these
> >> >>>>>>>>> define several runtime parameters, which in turn define how
> >> >>>>>>>>> well/badly
> >> >>>>>>>>> will the pipeline perform and how many resources might be
> >> >>>>>>>>> needed. From
> >> >>>>>>>>> my point of view, pure streaming should be the most resource
> >> >>>>>>>>> demanding
> >> >>>>>>>>> (if not, why bother with batch? why not run everything in
> >> >>>>>>>>> streaming
> >> >>>>>>>>> only? what will there remain to "unify"?).
> >> >>>>>>>>>
> >> >>>>>>>>> > Fortunately, for batch, only the state for a single key
> >> >>>>>>>>> needs to be
> >> >>>>>>>>> preserved at a time, rather than the state for all keys across
> >> >>>>>>>>> the range
> >> >>>>>>>>> of skew. Of course if you have few or hot keys, one can still
> >> >>>>>>>>> have
> >> >>>>>>>>> issues (and this is not specific to StatefulDoFns).
> >> >>>>>>>>>
> >> >>>>>>>>> Yes, but here is still the presumption that my stateful DoFn
> can
> >> >>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
> >> >>>>>>>>> case in
> >> >>>>>>>>> more detail.
> >> >>>>>>>>>
> >> >>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
> >> >>>>>>>>> some key for
> >> >>>>>>>>> each element, which is irrelevant for the demonstration). Your
> >> >>>>>>>>> task is
> >> >>>>>>>>> to calculate in running global window the actual number of
> >> >>>>>>>>> changes
> >> >>>>>>>>> between state 0 and state 1 and vice versa. When the state
> >> >>>>>>>>> doesn't
> >> >>>>>>>>> change, you don't calculate anything. If input (for given key)
> >> >>>>>>>>> would be
> >> >>>>>>>>> (tN denotes timestamp N):
> >> >>>>>>>>>
> >> >>>>>>>>> t1: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t2: 0
> >> >>>>>>>>>
> >> >>>>>>>>> t3: 0
> >> >>>>>>>>>
> >> >>>>>>>>> t4: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t5: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t6: 0
> >> >>>>>>>>>
> >> >>>>>>>>> then the output should yield (supposing that default state is
> >> >>>>>>>>> zero):
> >> >>>>>>>>>
> >> >>>>>>>>> t1: (one: 1, zero: 0)
> >> >>>>>>>>>
> >> >>>>>>>>> t2: (one: 1, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t3: (one: 1, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t4: (one: 2, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t5: (one: 2, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t6: (one: 2, zero: 2)
> >> >>>>>>>>>
> >> >>>>>>>>> How would you implement this in current Beam semantics?
> >> >>>>>>>> I think your saying here that I know that my input is ordered
> >> >>>>>>>> in a specific way and since I assume the order when writing my
> >> >>>>>>>> pipeline I can perform this optimization. But there is nothing
> >> >>>>>>>> preventing a runner from noticing that your processing in the
> >> >>>>>>>> global window with a specific type of trigger and re-ordering
> >> >>>>>>>> your inputs/processing to get better performance (since you
> >> >>>>>>>> can't use an AfterWatermark trigger for your pipeline in
> >> >>>>>>>> streaming for the GlobalWindow).
> >> >>>>>>>>
> >> >>>>>>>> Today, if you must have a strict order, you must guarantee that
> >> >>>>>>>> your StatefulParDo implements the necessary "buffering &
> >> >>>>>>>> sorting" into state. I can see why you would want an annotation
> >> >>>>>>>> that says I must have timestamp ordered elements, since it
> >> >>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
> >> >>>>>>>> is a low-level function, it really is the "here you go and do
> >> >>>>>>>> whatever you need to but here be dragons" function while
> >> >>>>>>>> windowing and triggering is meant to keep many people from
> >> >>>>>>>> writing StatefulParDo in the first place.
> >> >>>>>>>>
> >> >>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
> >> >>>>>>>>> are likely to
> >> >>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
> >> >>>>>>>>> falls
> >> >>>>>>>>> behind in streaming mode as well.
> >> >>>>>>>>>
> >> >>>>>>>>> But the worst case is defined by input of size (available
> >> >>>>>>>>> resources +
> >> >>>>>>>>> single byte) -> pipeline fail. Although it could have
> >> >>>>>>>>> finished, given
> >> >>>>>>>>> the right conditions.
> >> >>>>>>>>>
> >> >>>>>>>>> > This might be reasonable, implemented by default by
> >> >>>>>>>>> buffering
> >> >>>>>>>>> everything and releasing elements as the watermark (+lateness)
> >> >>>>>>>>> advances,
> >> >>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
> >> >>>>>>>>> reason
> >> >>>>>>>>> about) code.
> >> >>>>>>>>>
> >> >>>>>>>>> Sure, the pipeline will be less efficient, because it would
> >> >>>>>>>>> have to
> >> >>>>>>>>> buffer and sort the inputs. But at least it will produce
> >> >>>>>>>>> correct results
> >> >>>>>>>>> in cases where updates to state are order-sensitive.
> >> >>>>>>>>>
> >> >>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
> >> >>>>>>>>> (key, values):
> >> >>>>>>>>> [(key, value) for value in values])?
> >> >>>>>>>>>
> >> >>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
> >> >>>>>>>>> The trigger
> >> >>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
> >> >>>>>>>>> element
> >> >>>>>>>>> with lowest timestamp in the buffer. Although this could be
> >> >>>>>>>>> somehow
> >> >>>>>>>>> emulated by fixed trigger each X millis.
> >> >>>>>>>>>
> >> >>>>>>>>> > Or is the underlying desire just to be able to hint to
> >> >>>>>>>>> the runner
> >> >>>>>>>>> that the code may perform better (e.g. require less resources)
> >> >>>>>>>>> as skew
> >> >>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
> >> >>>>>>>>>
> >> >>>>>>>>> No, the sorting would have to be done in streaming case as
> >> >>>>>>>>> well. That is
> >> >>>>>>>>> an imperative of the unified model. I think it is possible to
> >> >>>>>>>>> sort by
> >> >>>>>>>>> timestamp only in batch case (and do it for *all* batch
> >> >>>>>>>>> stateful pardos
> >> >>>>>>>>> without annotation), or introduce annotation, but then make
> >> >>>>>>>>> the same
> >> >>>>>>>>> guarantees for streaming case as well.
> >> >>>>>>>>>
> >> >>>>>>>>> Jan
> >> >>>>>>>>>
> >> >>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
> >> >>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
> >> >>>>>>>>>> <je...@seznam.cz> wrote:
> >> >>>>>>>>>>> Hi Robert,
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
> >> >>>>>>>>>>> guarantees
> >> >>>>>>>>>>> of ordering are given in either mode, there is *implicit*
> >> >>>>>>>>>>> ordering in
> >> >>>>>>>>>>> streaming case that is due to nature of the processing - the
> >> >>>>>>>>>>> difference
> >> >>>>>>>>>>> between watermark and timestamp of elements flowing through
> >> >>>>>>>>>>> the pipeline
> >> >>>>>>>>>>> are generally low (too high difference leads to the
> >> >>>>>>>>>>> overbuffering
> >> >>>>>>>>>>> problem), but there is no such bound on batch.
> >> >>>>>>>>>> Fortunately, for batch, only the state for a single key needs
> >> >>>>>>>>>> to be
> >> >>>>>>>>>> preserved at a time, rather than the state for all keys
> >> >>>>>>>>>> across the
> >> >>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
> >> >>>>>>>>>> still
> >> >>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
> >> >>>>>>>>>>
> >> >>>>>>>>>>> As a result, I see a few possible solutions:
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> - the best and most natural seems to be extension of
> >> >>>>>>>>>>> the model, so
> >> >>>>>>>>>>> that it defines batch as not only "streaming pipeline
> >> >>>>>>>>>>> executed in batch
> >> >>>>>>>>>>> fashion", but "pipeline with at least as good runtime
> >> >>>>>>>>>>> characteristics as
> >> >>>>>>>>>>> in streaming case, executed in batch fashion", I really
> >> >>>>>>>>>>> don't think that
> >> >>>>>>>>>>> there are any conflicts with the current model, or that this
> >> >>>>>>>>>>> could
> >> >>>>>>>>>>> affect performance, because the required sorting (as
> pointed by
> >> >>>>>>>>>>> Aljoscha) is very probably already done during translation
> >> >>>>>>>>>>> of stateful
> >> >>>>>>>>>>> pardos. Also note that this definition only affects user
> >> >>>>>>>>>>> defined
> >> >>>>>>>>>>> stateful pardos
> >> >>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
> >> >>>>>>>>>> have
> >> >>>>>>>>>> microbatch, or even a runner that alternates between
> >> >>>>>>>>>> different modes.
> >> >>>>>>>>>> The model describes what the valid outputs are given a
> >> >>>>>>>>>> (sometimes
> >> >>>>>>>>>> partial) set of inputs. It becomes really hard to define
> >> >>>>>>>>>> things like
> >> >>>>>>>>>> "as good runtime characteristics." Once you allow any
> >> >>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
> >> >>>>>>>>>> (and
> >> >>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
> >> >>>>>>>>>> out-of-orderedness.
> >> >>>>>>>>>>
> >> >>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
> >> >>>>>>>>>> likely to
> >> >>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
> >> >>>>>>>>>> falls
> >> >>>>>>>>>> behind in streaming mode as well.
> >> >>>>>>>>>>
> >> >>>>>>>>>>> - another option would be to introduce annotation for
> >> >>>>>>>>>>> DoFns (e.g.
> >> >>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
> >> >>>>>>>>>>> the sorting
> >> >>>>>>>>>>> in batch case - but - this extension would have to ensure
> >> >>>>>>>>>>> the sorting in
> >> >>>>>>>>>>> streaming mode also - it would require definition of allowed
> >> >>>>>>>>>>> lateness,
> >> >>>>>>>>>>> and triggger (essentially similar to window)
> >> >>>>>>>>>> This might be reasonable, implemented by default by buffering
> >> >>>>>>>>>> everything and releasing elements as the watermark
> (+lateness)
> >> >>>>>>>>>> advances, but would likely lead to inefficient (though
> >> >>>>>>>>>> *maybe* easier
> >> >>>>>>>>>> to reason about) code. Not sure about the semantics of
> >> >>>>>>>>>> triggering
> >> >>>>>>>>>> here, especially data-driven triggers. Would it be roughly
> >> >>>>>>>>>> equivalent
> >> >>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
> >> >>>>>>>>>> value in
> >> >>>>>>>>>> values])?
> >> >>>>>>>>>>
> >> >>>>>>>>>> Or is the underlying desire just to be able to hint to the
> >> >>>>>>>>>> runner that
> >> >>>>>>>>>> the code may perform better (e.g. require less resources) as
> >> >>>>>>>>>> skew is
> >> >>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
> >> >>>>>>>>>>
> >> >>>>>>>>>>> - last option would be to introduce these "higher order
> >> >>>>>>>>>>> guarantees" in
> >> >>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
> >> >>>>>>>>>>> worst
> >> >>>>>>>>>>> option to me
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> I see the first two options quite equally good, although the
> >> >>>>>>>>>>> letter one
> >> >>>>>>>>>>> is probably more time consuming to implement. But it would
> >> >>>>>>>>>>> bring
> >> >>>>>>>>>>> additional feature to streaming case as well.
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> Thanks for any thoughts.
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> Jan
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
> >> >>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
> >> >>>>>>>>>>>> <je...@seznam.cz> wrote:
> >> >>>>>>>>>>>>> Hi Reuven,
> >> >>>>>>>>>>>>>
> >> >>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
> >> >>>>>>>>>>>>>> runners.
> >> >>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
> >> >>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
> >> >>>>>>>>>>>>> of elements. That basically (practically) can work only
> >> >>>>>>>>>>>>> for cases, where the order of input elements doesn't
> >> >>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
> >> >>>>>>>>>>>>> time you have a state machine involved, then the ordering
> >> >>>>>>>>>>>>> of elements would matter.
> >> >>>>>>>>>>>> No guarantees on order are provided in *either* streaming
> >> >>>>>>>>>>>> or batch
> >> >>>>>>>>>>>> mode by the model. However, it is the case that in order to
> >> >>>>>>>>>>>> make
> >> >>>>>>>>>>>> forward progress most streaming runners attempt to limit
> >> >>>>>>>>>>>> the amount of
> >> >>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
> >> >>>>>>>>>>>> processing
> >> >>>>>>>>>>>> time) to make forward progress, which in turn could help
> >> >>>>>>>>>>>> cap the
> >> >>>>>>>>>>>> amount of state that must be held concurrently, whereas a
> >> >>>>>>>>>>>> batch runner
> >> >>>>>>>>>>>> may not allow any state to be safely discarded until the
> whole
> >> >>>>>>>>>>>> timeline from infinite past to infinite future has been
> >> >>>>>>>>>>>> observed.
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
> >> >>>>>>>>>>>> batch" in batch mode.
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
> >> >>>>>>>>>>>> <mx...@apache.org> wrote:
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>>>> batch semantics and streaming semantics differs only
> >> >>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
> >> >>>>>>>>>>>>>> batch and cannot on stream
> >> >>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
> >> >>>>>>>>>>>>> trigger. You
> >> >>>>>>>>>>>>> could define additional triggers that do early firings.
> >> >>>>>>>>>>>>> And you could
> >> >>>>>>>>>>>>> even trigger the global window by advancing the watermark
> >> >>>>>>>>>>>>> to +inf.
> >> >>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
> >> >>>>>>>>>>>> default
> >> >>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
> >> >>>>>>>>>>>> is more
> >> >>>>>>>>>>>> likely to be user error than an actual desire to have no
> >> >>>>>>>>>>>> output until
> >> >>>>>>>>>>>> drain. But it's semantically valid in the model.
>
>
Re: Definition of Unified model
Posted by Jan Lukavský <je...@seznam.cz>.
Hi,
yes. It seems that ordering by user supplied UDF makes sense and I will
update the design proposal accordingly.
Would that solve the issues you mention?
Jan
---------- Původní e-mail ----------
Od: Reuven Lax <re...@google.com>
Komu: dev <de...@beam.apache.org>
Datum: 23. 5. 2019 18:44:38
Předmět: Re: Definition of Unified model
"
I'm simply saying that timestamp ordering is insufficient for state
machines. I wasn't proposing Kafka as a solution - that was simply an
example of how people solve this problem in other scenarios.
BTW another example of ordering: Imagine today that you have a triggered Sum
aggregation writing out to a key-value sink. In theory we provide no
ordering, so the sink might write the triggered sums in the wrong order,
ending up with an incorrect value in the sink. In this case you probably
want values ordered by trigger pane index.
Reuven
On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)> wrote:
"
Hi Reuven,
I share the view point of Robert. I think the isuue you refer to is not in
reality related to timestamps, but to the fact, that ordering of events in
time is observer dependent (either caused by relativity, or time skew,
essentially this has the same consequences). And the resolution in fact isn'
t Kafka, but generally an authoritative observer, that tells you "I saw the
events in this order". And you either have one (and have the outcome of his
observation persisted in the data - e.g. as offset in Kafka partition), then
you should be able to use it (maybe that suggests afterall that sorting by
some user supplied UDF might make sense), or do not have it, and then any
interpretation of the data seems to be equally valid. Although determinism
is fine, of course.
Jan
---------- Původní e-mail ----------
Od: Reuven Lax <relax@google.com(mailto:relax@google.com)>
Komu: dev <dev@beam.apache.org(mailto:dev@beam.apache.org)>
Datum: 23. 5. 2019 17:39:12
Předmět: Re: Definition of Unified model
"
So an example would be elements of type "startUserSession" and
"endUserSession" (website sessions, not Beam sessions). Logically you may
need to process them in the correct order if you have any sort of state-
machine logic. However timestamp ordering is never guaranteed to match the
logical ordering. Not only might you have several elements with the same
timestamp, but in reality time skew across backend servers can cause the
events to have timestamps in reverse order of the actual causality order.
People do solve this problem today though. Publish the events to Kafka,
making sure that events for the same user end up in the same Kafka
partition. This ensures that the events appear in the Kafka partitions in
causality order, even if the timestamp order doesn't match. The your Kafka
subscriber simply process the elements in each partition in order.
I think the ability to impose FIFO causality ordering is what's needed for
any state-machine work. Timestamp ordering has advantages (though often I
think the advantage is in state), but does not solve this problem.
Reuven
On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <robertwb@google.com
(mailto:robertwb@google.com)> wrote:
"Good point.
The "implementation-specific" way I would do this is
window-by-instant, followed by a DoFn that gets all the elements with
the same timestamp and sorts/acts accordingly, but this counts on the
runner producing windows in timestamp order (likely?) and also the
subsequent DoFn getting them in this order (also likely, due to
fusion).
One could make the argument that, though it does not provide
deterministic behavior, getting elements of the same timestamp in
different orders should produce equally valid interpretations of the
data. (After all, due to relatively, timestamps are not technically
well ordered across space.) I can see how data-dependent tiebreakers
could be useful, or promises of preservation of order between
operations.
- Robert
On Thu, May 23, 2019 at 4:18 PM Reuven Lax <relax@google.com
(mailto:relax@google.com)> wrote:
>
> So Jan's example of state machines is quite a valid use case for ordering.
However in my experience, timestamp ordering is insufficient for state
machines. Elements that cause state transitions might come in with the exact
same timestamp, yet still have a necessary ordering. Especially given Beam's
decision to have milliseconds timestamps this is possible, but even at
microsecond or nanosecond precision this can happen at scale. To handle
state machines you usually need some sort of FIFO ordering along with an
ordered sources, such as Kafka, not timestamp ordering.
>
> Reuven
>
> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)> wrote:
>>
>> Hi all,
>>
>> thanks everyone for this discussion. I think I have gathered enough
>> feedback to be able to put down a proposition for changes, which I will
>> do and send to this list for further discussion. There are still doubts
>> remaining the non-determinism and it's relation to outputs stability vs.
>> latency. But I will try to clarify all this in the design document.
>>
>> Thanks,
>>
>> Jan
>>
>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>> >> Someone from Flink might correct me if I'm wrong, but that's my
>> >> current understanding.
>> >
>> > In essence your description of how exactly-once works in Flink is
>> > correct. The general assumption in Flink is that pipelines must be
>> > deterministic and thus produce idempotent writes in the case of
>> > failures. However, that doesn't mean Beam sinks can't guarantee a bit
>> > more with what Flink has to offer.
>> >
>> > Luke already mentioned the design discussions for @RequiresStableInput
>> > which ensures idempotent writes for non-deterministic pipelines. This
>> > is not part of the model but an optional Beam feature.
>> >
>> > We recently implemented support for @RequiresStableInput in the Flink
>> > Runner. Reuven mentioned the Flink checkpoint confirmation, which
>> > allows us to buffer (and checkpoint) processed data and only emit it
>> > once a Flink checkpoint has completed.
>> >
>> > Cheers,
>> > Max
>> >
>> > On 21.05.19 16:49, Jan Lukavský wrote:
>> >> Hi,
>> >>
>> >> > Actually, I think it is a larger (open) question whether exactly
>> >> once is guaranteed by the model or whether runners are allowed to
>> >> relax that. I would think, however, that sources correctly
>> >> implemented should be idempotent when run atop an exactly once
>> >> infrastructure such as Flink of Dataflow.
>> >>
>> >> I would assume, that the model basically inherits guarantees of
>> >> underlying infrastructure. Because Flink does not work as you
>> >> described (atomic commit of inputs, state and outputs), but rather a
>> >> checkpoint mark is flowing through the DAG much like watermark and on
>> >> failures operators are restored and data reprocessed, it (IMHO)
>> >> implies, that you have exactly once everywhere in the DAG *but*
>> >> sinks. That is because sinks cannot be restored to previous state,
>> >> instead sinks are supposed to be idempotent in order for the exactly
>> >> once to really work (or at least be able to commit outputs on
>> >> checkpoint in sink). That implies that if you don't have sink that is
>> >> able to commit outputs atomically on checkpoint, the pipeline
>> >> execution should be deterministic upon retries, otherwise shadow
>> >> writes from failed paths of the pipeline might appear.
>> >>
>> >> Someone from Flink might correct me if I'm wrong, but that's my
>> >> current understanding.
>> >>
>> >> > Sounds like we should make this clearer.
>> >>
>> >> I meant that you are right that we must not in any thoughts we are
>> >> having forget that streams are by definition out-of-order. That is
>> >> property that we cannot change. But - that doesn't limit us from
>> >> creating operator that presents the data to UDF as if the stream was
>> >> ideally sorted. It can do that by introducing latency, of course.
>> >>
>> >> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>> >>> Reza: One could provide something like this as a utility class, but
>> >>> one downside is that it is not scale invariant. It requires a tuning
>> >>> parameter that, if to small, won't mitigate the problem, but if to
>> >>> big, greatly increases latency. (Possibly one could define a dynamic
>> >>> session-like window to solve this though...) It also might be harder
>> >>> for runners that *can* cheaply present stuff in timestamp order to
>> >>> optimize. (That and, in practice, our annotation-style process
methods
>> >>> don't lend themselves to easy composition.) I think it could work in
>> >>> specific cases though.
>> >>>
>> >>> More inline below.
>> >>>
>> >>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)> wrote:
>> >>>> Hi Robert,
>> >>>>
>> >>>> > Beam has an exactly-once model. If the data was consumed, state
>> >>>> mutated, and outputs written downstream (these three are committed
>> >>>> together atomically) it will not be replayed. That does not, of
>> >>>> course,
>> >>>> solve the non-determanism due to ordering (including the fact that
two
>> >>>> operations reading the same PCollection may view different
ordering).
>> >>>>
>> >>>> I think what you describe is a property of a runner, not of the
model,
>> >>>> right? I think if I run my pipeline on Flink I will not get this
>> >>>> atomicity, because although Flink uses also exactly-once model if
>> >>>> might
>> >>>> write outputs multiple times.
>> >>> Actually, I think it is a larger (open) question whether exactly once
>> >>> is guaranteed by the model or whether runners are allowed to relax
>> >>> that. I would think, however, that sources correctly implemented
>> >>> should be idempotent when run atop an exactly once infrastructure
such
>> >>> as Flink of Dataflow.
>> >>>
>> >>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
>> >>>> received
>> >>>> in a specific order? In the current model, it is not. Being able to
>> >>>> read, handle, and produced out-of-order data, including late data,
>> >>>> is a
>> >>>> pretty fundamental property of distributed systems.
>> >>>>
>> >>>> Yes, absolutely. The argument here is not that Stateful ParDo should
>> >>>> presume to receive elements in any order, but to _present_ it as
>> >>>> such to
>> >>>> the user @ProcessElement function.
>> >>> Sounds like we should make this clearer.
>> >>>
>> >>>> > 2) Given that some operations are easier (or possibly only
>> >>>> possible)
>> >>>> to write when operating on ordered data, and that different runners
>> >>>> may
>> >>>> have (significantly) cheaper ways to provide this ordering than can
be
>> >>>> done by the user themselves, should we elevate this to a property of
>> >>>> (Stateful?)DoFns that the runner can provide? I think a compelling
>> >>>> argument can be made here that we should.
>> >>>>
>> >>>> +1
>> >>>>
>> >>>> Jan
>> >>>>
>> >>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>> >>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)> wrote:
>> >>>>>> > I don't see batch vs. streaming as part of the model. One
>> >>>>>> can have
>> >>>>>> microbatch, or even a runner that alternates between different
>> >>>>>> modes.
>> >>>>>>
>> >>>>>> Although I understand motivation of this statement, this project
>> >>>>>> name is
>> >>>>>> "Apache Beam: An advanced unified programming model". What does
the
>> >>>>>> model unify, if "streaming vs. batch" is not part of the model?
>> >>>>> What I mean is that streaming vs. batch is no longer part of the
>> >>>>> model
>> >>>>> (or ideally API), but pushed down to be a concern of the runner
>> >>>>> (executor) of the pipeline.
>> >>>>>
>> >>>>>
>> >>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)>
>> >>>>> wrote:
>> >>>>>> Hi Kenn,
>> >>>>>>
>> >>>>>> OK, so if we introduce annotation, we can have stateful ParDo
>> >>>>>> with sorting, that would perfectly resolve my issues. I still
>> >>>>>> have some doubts, though. Let me explain. The current behavior of
>> >>>>>> stateful ParDo has the following properties:
>> >>>>>>
>> >>>>>> a) might fail in batch, although runs fine in streaming (that
>> >>>>>> is due to the buffering, and unbounded lateness in batch, which
>> >>>>>> was discussed back and forth in this thread)
>> >>>>>>
>> >>>>>> b) might be non deterministic (this is because the elements
>> >>>>>> arrive at somewhat random order, and even if you do the operation
>> >>>>>> "assign unique ID to elements" this might produce different
>> >>>>>> results when run multiple times)
>> >>>>> PCollections are *explicitly* unordered. Any operations that
>> >>>>> assume or
>> >>>>> depend on a specific ordering for correctness (or determinism) must
>> >>>>> provide that ordering themselves (i.e. tolerate "arbitrary
shuffling
>> >>>>> of inputs"). As you point out, that may be very expensive if you
have
>> >>>>> very hot keys with very large (unbounded) timestamp skew.
>> >>>>>
>> >>>>> StatefulDoFns are low-level operations that should be used with
care;
>> >>>>> the simpler windowing model gives determinism in the face of
>> >>>>> unordered
>> >>>>> data (though late data and non-end-of-window triggering introduces
>> >>>>> some of the non-determanism back in).
>> >>>>>
>> >>>>>> What worries me most is the property b), because it seems to me
>> >>>>>> to have serious consequences - not only that if you run twice
>> >>>>>> batch pipeline you would get different results, but even on
>> >>>>>> streaming, when pipeline fails and gets restarted from
>> >>>>>> checkpoint, produced output might differ from the previous run
>> >>>>>> and data from the first run might have already been persisted
>> >>>>>> into sink. That would create somewhat messy outputs.
>> >>>>> Beam has an exactly-once model. If the data was consumed, state
>> >>>>> mutated, and outputs written downstream (these three are committed
>> >>>>> together atomically) it will not be replayed. That does not, of
>> >>>>> course, solve the non-determanism due to ordering (including the
fact
>> >>>>> that two operations reading the same PCollection may view different
>> >>>>> ordering).
>> >>>>>
>> >>>>>> These two properties makes me think that the current
>> >>>>>> implementation is more of a _special case_ than the general one.
>> >>>>>> The general one would be that your state doesn't have the
>> >>>>>> properties to be able to tolerate buffering problems and/or
>> >>>>>> non-determinism. Which is the case where you need sorting in both
>> >>>>>> streaming and batch to be part of the model.
>> >>>>>>
>> >>>>>> Let me point out one more analogy - that is merging vs.
>> >>>>>> non-merging windows. The general case (merging windows) implies
>> >>>>>> sorting by timestamp in both batch case (explicit) and streaming
>> >>>>>> (buffering). The special case (non-merging windows) doesn't rely
>> >>>>>> on any timestamp ordering, so the sorting and buffering can be
>> >>>>>> dropped. The underlying root cause of this is the same for both
>> >>>>>> stateful ParDo and windowing (essentially, assigning window
>> >>>>>> labels is a stateful operation when windowing function is
merging).
>> >>>>>>
>> >>>>>> The reason for the current behavior of stateful ParDo seems to be
>> >>>>>> performance, but is it right to abandon correctness in favor of
>> >>>>>> performance? Wouldn't it be more consistent to have the default
>> >>>>>> behavior prefer correctness and when you have the specific
>> >>>>>> conditions of state function having special properties, then you
>> >>>>>> can annotate your DoFn (with something like
>> >>>>>> @TimeOrderingAgnostic), which would yield a better performance in
>> >>>>>> that case?
>> >>>>> There are two separable questions here.
>> >>>>>
>> >>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are
received
>> >>>>> in a specific order? In the current model, it is not. Being able to
>> >>>>> read, handle, and produced out-of-order data, including late data,
is
>> >>>>> a pretty fundamental property of distributed systems.
>> >>>>>
>> >>>>> 2) Given that some operations are easier (or possibly only
possible)
>> >>>>> to write when operating on ordered data, and that different runners
>> >>>>> may have (significantly) cheaper ways to provide this ordering than
>> >>>>> can be done by the user themselves, should we elevate this to a
>> >>>>> property of (Stateful?)DoFns that the runner can provide? I think a
>> >>>>> compelling argument can be made here that we should.
>> >>>>>
>> >>>>> - Robert
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>> >>>>>>
>> >>>>>> Thanks for the nice small example of a calculation that depends
>> >>>>>> on order. You are right that many state machines have this
>> >>>>>> property. I agree w/ you and Luke that it is convenient for batch
>> >>>>>> processing to sort by event timestamp before running a stateful
>> >>>>>> ParDo. In streaming you could also implement "sort by event
>> >>>>>> timestamp" by buffering until you know all earlier data will be
>> >>>>>> dropped - a slack buffer up to allowed lateness.
>> >>>>>>
>> >>>>>> I do not think that it is OK to sort in batch and not in
>> >>>>>> streaming. Many state machines diverge very rapidly when things
>> >>>>>> are out of order. So each runner if they see the
>> >>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
>> >>>>>> sorted data (by some mix of buffering and dropping), or to reject
>> >>>>>> the pipeline as unsupported.
>> >>>>>>
>> >>>>>> And also want to say that this is not the default case - many
>> >>>>>> uses of state & timers in ParDo yield different results at the
>> >>>>>> element level, but the results are equivalent at in the big
>> >>>>>> picture. Such as the example of "assign a unique sequence number
>> >>>>>> to each element" or "group into batches" it doesn't matter
>> >>>>>> exactly what the result is, only that it meets the spec. And
>> >>>>>> other cases like user funnels are monotonic enough that you also
>> >>>>>> don't actually need sorting.
>> >>>>>>
>> >>>>>> Kenn
>> >>>>>>
>> >>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)>
>> >>>>>> wrote:
>> >>>>>>> Yes, the problem will arise probably mostly when you have not
>> >>>>>>> well distributed keys (or too few keys). I'm really not sure if
>> >>>>>>> a pure GBK with a trigger can solve this - it might help to have
>> >>>>>>> data driven trigger. There would still be some doubts, though.
>> >>>>>>> The main question is still here - people say, that sorting by
>> >>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
>> >>>>>>> I don't really see why - the sorting is very probably already
>> >>>>>>> there. And if not (hash grouping instead of sorted grouping),
>> >>>>>>> then the sorting would affect only user defined StatefulParDos.
>> >>>>>>>
>> >>>>>>> This would suggest that the best way out of this would be really
>> >>>>>>> to add annotation, so that the author of the pipeline can decide.
>> >>>>>>>
>> >>>>>>> If that would be acceptable I think I can try to prepare some
>> >>>>>>> basic functionality, but I'm not sure, if I would be able to
>> >>>>>>> cover all runners / sdks.
>> >>>>>>>
>> >>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>> >>>>>>>
>> >>>>>>> It is read all per key and window and not just read all (this
>> >>>>>>> still won't scale with hot keys in the global window). The GBK
>> >>>>>>> preceding the StatefulParDo will guarantee that you are
>> >>>>>>> processing all the values for a specific key and window at any
>> >>>>>>> given time. Is there a specific window/trigger that is missing
>> >>>>>>> that you feel would remove the need for you to use StatefulParDo?
>> >>>>>>>
>> >>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)>
>> >>>>>>> wrote:
>> >>>>>>>> Hi Lukasz,
>> >>>>>>>>
>> >>>>>>>>> Today, if you must have a strict order, you must guarantee
>> >>>>>>>>> that your StatefulParDo implements the necessary "buffering &
>> >>>>>>>>> sorting" into state.
>> >>>>>>>> Yes, no problem with that. But this whole discussion started,
>> >>>>>>>> because *this doesn't work on batch*. You simply cannot first
>> >>>>>>>> read everything from distributed storage and then buffer it all
>> >>>>>>>> into memory, just to read it again, but sorted. That will not
>> >>>>>>>> work. And even if it would, it would be a terrible waste of
>> >>>>>>>> resources.
>> >>>>>>>>
>> >>>>>>>> Jan
>> >>>>>>>>
>> >>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>> >>>>>>>>
>> >>>>>>>>
>> >>>>>>>>
>> >>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)>
>> >>>>>>>> wrote:
>> >>>>>>>>> This discussion brings many really interesting questions for
>> >>>>>>>>> me. :-)
>> >>>>>>>>>
>> >>>>>>>>> > I don't see batch vs. streaming as part of the model. One
>> >>>>>>>>> can have
>> >>>>>>>>> microbatch, or even a runner that alternates between different
>> >>>>>>>>> modes.
>> >>>>>>>>>
>> >>>>>>>>> Although I understand motivation of this statement, this
>> >>>>>>>>> project name is
>> >>>>>>>>> "Apache Beam: An advanced unified programming model". What
>> >>>>>>>>> does the
>> >>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>> >>>>>>>>>
>> >>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
>> >>>>>>>>> are
>> >>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
>> >>>>>>>>> All these
>> >>>>>>>>> define several runtime parameters, which in turn define how
>> >>>>>>>>> well/badly
>> >>>>>>>>> will the pipeline perform and how many resources might be
>> >>>>>>>>> needed. From
>> >>>>>>>>> my point of view, pure streaming should be the most resource
>> >>>>>>>>> demanding
>> >>>>>>>>> (if not, why bother with batch? why not run everything in
>> >>>>>>>>> streaming
>> >>>>>>>>> only? what will there remain to "unify"?).
>> >>>>>>>>>
>> >>>>>>>>> > Fortunately, for batch, only the state for a single key
>> >>>>>>>>> needs to be
>> >>>>>>>>> preserved at a time, rather than the state for all keys across
>> >>>>>>>>> the range
>> >>>>>>>>> of skew. Of course if you have few or hot keys, one can still
>> >>>>>>>>> have
>> >>>>>>>>> issues (and this is not specific to StatefulDoFns).
>> >>>>>>>>>
>> >>>>>>>>> Yes, but here is still the presumption that my stateful DoFn
can
>> >>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
>> >>>>>>>>> case in
>> >>>>>>>>> more detail.
>> >>>>>>>>>
>> >>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
>> >>>>>>>>> some key for
>> >>>>>>>>> each element, which is irrelevant for the demonstration). Your
>> >>>>>>>>> task is
>> >>>>>>>>> to calculate in running global window the actual number of
>> >>>>>>>>> changes
>> >>>>>>>>> between state 0 and state 1 and vice versa. When the state
>> >>>>>>>>> doesn't
>> >>>>>>>>> change, you don't calculate anything. If input (for given key)
>> >>>>>>>>> would be
>> >>>>>>>>> (tN denotes timestamp N):
>> >>>>>>>>>
>> >>>>>>>>> t1: 1
>> >>>>>>>>>
>> >>>>>>>>> t2: 0
>> >>>>>>>>>
>> >>>>>>>>> t3: 0
>> >>>>>>>>>
>> >>>>>>>>> t4: 1
>> >>>>>>>>>
>> >>>>>>>>> t5: 1
>> >>>>>>>>>
>> >>>>>>>>> t6: 0
>> >>>>>>>>>
>> >>>>>>>>> then the output should yield (supposing that default state is
>> >>>>>>>>> zero):
>> >>>>>>>>>
>> >>>>>>>>> t1: (one: 1, zero: 0)
>> >>>>>>>>>
>> >>>>>>>>> t2: (one: 1, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t3: (one: 1, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t4: (one: 2, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t5: (one: 2, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t6: (one: 2, zero: 2)
>> >>>>>>>>>
>> >>>>>>>>> How would you implement this in current Beam semantics?
>> >>>>>>>> I think your saying here that I know that my input is ordered
>> >>>>>>>> in a specific way and since I assume the order when writing my
>> >>>>>>>> pipeline I can perform this optimization. But there is nothing
>> >>>>>>>> preventing a runner from noticing that your processing in the
>> >>>>>>>> global window with a specific type of trigger and re-ordering
>> >>>>>>>> your inputs/processing to get better performance (since you
>> >>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>> >>>>>>>> streaming for the GlobalWindow).
>> >>>>>>>>
>> >>>>>>>> Today, if you must have a strict order, you must guarantee that
>> >>>>>>>> your StatefulParDo implements the necessary "buffering &
>> >>>>>>>> sorting" into state. I can see why you would want an annotation
>> >>>>>>>> that says I must have timestamp ordered elements, since it
>> >>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
>> >>>>>>>> is a low-level function, it really is the "here you go and do
>> >>>>>>>> whatever you need to but here be dragons" function while
>> >>>>>>>> windowing and triggering is meant to keep many people from
>> >>>>>>>> writing StatefulParDo in the first place.
>> >>>>>>>>
>> >>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
>> >>>>>>>>> are likely to
>> >>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>> >>>>>>>>> falls
>> >>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>
>> >>>>>>>>> But the worst case is defined by input of size (available
>> >>>>>>>>> resources +
>> >>>>>>>>> single byte) -> pipeline fail. Although it could have
>> >>>>>>>>> finished, given
>> >>>>>>>>> the right conditions.
>> >>>>>>>>>
>> >>>>>>>>> > This might be reasonable, implemented by default by
>> >>>>>>>>> buffering
>> >>>>>>>>> everything and releasing elements as the watermark (+lateness)
>> >>>>>>>>> advances,
>> >>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
>> >>>>>>>>> reason
>> >>>>>>>>> about) code.
>> >>>>>>>>>
>> >>>>>>>>> Sure, the pipeline will be less efficient, because it would
>> >>>>>>>>> have to
>> >>>>>>>>> buffer and sort the inputs. But at least it will produce
>> >>>>>>>>> correct results
>> >>>>>>>>> in cases where updates to state are order-sensitive.
>> >>>>>>>>>
>> >>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
>> >>>>>>>>> (key, values):
>> >>>>>>>>> [(key, value) for value in values])?
>> >>>>>>>>>
>> >>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
>> >>>>>>>>> The trigger
>> >>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
>> >>>>>>>>> element
>> >>>>>>>>> with lowest timestamp in the buffer. Although this could be
>> >>>>>>>>> somehow
>> >>>>>>>>> emulated by fixed trigger each X millis.
>> >>>>>>>>>
>> >>>>>>>>> > Or is the underlying desire just to be able to hint to
>> >>>>>>>>> the runner
>> >>>>>>>>> that the code may perform better (e.g. require less resources)
>> >>>>>>>>> as skew
>> >>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
>> >>>>>>>>>
>> >>>>>>>>> No, the sorting would have to be done in streaming case as
>> >>>>>>>>> well. That is
>> >>>>>>>>> an imperative of the unified model. I think it is possible to
>> >>>>>>>>> sort by
>> >>>>>>>>> timestamp only in batch case (and do it for *all* batch
>> >>>>>>>>> stateful pardos
>> >>>>>>>>> without annotation), or introduce annotation, but then make
>> >>>>>>>>> the same
>> >>>>>>>>> guarantees for streaming case as well.
>> >>>>>>>>>
>> >>>>>>>>> Jan
>> >>>>>>>>>
>> >>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>> >>>>>>>>>> <je.ik@seznam.cz(mailto:je.ik@seznam.cz)> wrote:
>> >>>>>>>>>>> Hi Robert,
>> >>>>>>>>>>>
>> >>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
>> >>>>>>>>>>> guarantees
>> >>>>>>>>>>> of ordering are given in either mode, there is *implicit*
>> >>>>>>>>>>> ordering in
>> >>>>>>>>>>> streaming case that is due to nature of the processing - the
>> >>>>>>>>>>> difference
>> >>>>>>>>>>> between watermark and timestamp of elements flowing through
>> >>>>>>>>>>> the pipeline
>> >>>>>>>>>>> are generally low (too high difference leads to the
>> >>>>>>>>>>> overbuffering
>> >>>>>>>>>>> problem), but there is no such bound on batch.
>> >>>>>>>>>> Fortunately, for batch, only the state for a single key needs
>> >>>>>>>>>> to be
>> >>>>>>>>>> preserved at a time, rather than the state for all keys
>> >>>>>>>>>> across the
>> >>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
>> >>>>>>>>>> still
>> >>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>> >>>>>>>>>>
>> >>>>>>>>>>> As a result, I see a few possible solutions:
>> >>>>>>>>>>>
>> >>>>>>>>>>> - the best and most natural seems to be extension of
>> >>>>>>>>>>> the model, so
>> >>>>>>>>>>> that it defines batch as not only "streaming pipeline
>> >>>>>>>>>>> executed in batch
>> >>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>> >>>>>>>>>>> characteristics as
>> >>>>>>>>>>> in streaming case, executed in batch fashion", I really
>> >>>>>>>>>>> don't think that
>> >>>>>>>>>>> there are any conflicts with the current model, or that this
>> >>>>>>>>>>> could
>> >>>>>>>>>>> affect performance, because the required sorting (as pointed
by
>> >>>>>>>>>>> Aljoscha) is very probably already done during translation
>> >>>>>>>>>>> of stateful
>> >>>>>>>>>>> pardos. Also note that this definition only affects user
>> >>>>>>>>>>> defined
>> >>>>>>>>>>> stateful pardos
>> >>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
>> >>>>>>>>>> have
>> >>>>>>>>>> microbatch, or even a runner that alternates between
>> >>>>>>>>>> different modes.
>> >>>>>>>>>> The model describes what the valid outputs are given a
>> >>>>>>>>>> (sometimes
>> >>>>>>>>>> partial) set of inputs. It becomes really hard to define
>> >>>>>>>>>> things like
>> >>>>>>>>>> "as good runtime characteristics." Once you allow any
>> >>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
>> >>>>>>>>>> (and
>> >>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>> >>>>>>>>>> out-of-orderedness.
>> >>>>>>>>>>
>> >>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
>> >>>>>>>>>> likely to
>> >>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>> >>>>>>>>>> falls
>> >>>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>>
>> >>>>>>>>>>> - another option would be to introduce annotation for
>> >>>>>>>>>>> DoFns (e.g.
>> >>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
>> >>>>>>>>>>> the sorting
>> >>>>>>>>>>> in batch case - but - this extension would have to ensure
>> >>>>>>>>>>> the sorting in
>> >>>>>>>>>>> streaming mode also - it would require definition of allowed
>> >>>>>>>>>>> lateness,
>> >>>>>>>>>>> and triggger (essentially similar to window)
>> >>>>>>>>>> This might be reasonable, implemented by default by buffering
>> >>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>> >>>>>>>>>> advances, but would likely lead to inefficient (though
>> >>>>>>>>>> *maybe* easier
>> >>>>>>>>>> to reason about) code. Not sure about the semantics of
>> >>>>>>>>>> triggering
>> >>>>>>>>>> here, especially data-driven triggers. Would it be roughly
>> >>>>>>>>>> equivalent
>> >>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
>> >>>>>>>>>> value in
>> >>>>>>>>>> values])?
>> >>>>>>>>>>
>> >>>>>>>>>> Or is the underlying desire just to be able to hint to the
>> >>>>>>>>>> runner that
>> >>>>>>>>>> the code may perform better (e.g. require less resources) as
>> >>>>>>>>>> skew is
>> >>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>> >>>>>>>>>>
>> >>>>>>>>>>> - last option would be to introduce these "higher order
>> >>>>>>>>>>> guarantees" in
>> >>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
>> >>>>>>>>>>> worst
>> >>>>>>>>>>> option to me
>> >>>>>>>>>>>
>> >>>>>>>>>>> I see the first two options quite equally good, although the
>> >>>>>>>>>>> letter one
>> >>>>>>>>>>> is probably more time consuming to implement. But it would
>> >>>>>>>>>>> bring
>> >>>>>>>>>>> additional feature to streaming case as well.
>> >>>>>>>>>>>
>> >>>>>>>>>>> Thanks for any thoughts.
>> >>>>>>>>>>>
>> >>>>>>>>>>> Jan
>> >>>>>>>>>>>
>> >>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>> >>>>>>>>>>>> <je.ik@seznam.cz(mailto:je.ik@seznam.cz)> wrote:
>> >>>>>>>>>>>>> Hi Reuven,
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
>> >>>>>>>>>>>>>> runners.
>> >>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
>> >>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
>> >>>>>>>>>>>>> of elements. That basically (practically) can work only
>> >>>>>>>>>>>>> for cases, where the order of input elements doesn't
>> >>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
>> >>>>>>>>>>>>> time you have a state machine involved, then the ordering
>> >>>>>>>>>>>>> of elements would matter.
>> >>>>>>>>>>>> No guarantees on order are provided in *either* streaming
>> >>>>>>>>>>>> or batch
>> >>>>>>>>>>>> mode by the model. However, it is the case that in order to
>> >>>>>>>>>>>> make
>> >>>>>>>>>>>> forward progress most streaming runners attempt to limit
>> >>>>>>>>>>>> the amount of
>> >>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
>> >>>>>>>>>>>> processing
>> >>>>>>>>>>>> time) to make forward progress, which in turn could help
>> >>>>>>>>>>>> cap the
>> >>>>>>>>>>>> amount of state that must be held concurrently, whereas a
>> >>>>>>>>>>>> batch runner
>> >>>>>>>>>>>> may not allow any state to be safely discarded until the
whole
>> >>>>>>>>>>>> timeline from infinite past to infinite future has been
>> >>>>>>>>>>>> observed.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
>> >>>>>>>>>>>> batch" in batch mode.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>> >>>>>>>>>>>> <mxm@apache.org(mailto:mxm@apache.org)> wrote:
>> >>>>>>>>>>>>
>> >>>>>>>>>>>>>> batch semantics and streaming semantics differs only
>> >>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
>> >>>>>>>>>>>>>> batch and cannot on stream
>> >>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
>> >>>>>>>>>>>>> trigger. You
>> >>>>>>>>>>>>> could define additional triggers that do early firings.
>> >>>>>>>>>>>>> And you could
>> >>>>>>>>>>>>> even trigger the global window by advancing the watermark
>> >>>>>>>>>>>>> to +inf.
>> >>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
>> >>>>>>>>>>>> default
>> >>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
>> >>>>>>>>>>>> is more
>> >>>>>>>>>>>> likely to be user error than an actual desire to have no
>> >>>>>>>>>>>> output until
>> >>>>>>>>>>>> drain. But it's semantically valid in the model.
"
"
"
"
Re: Definition of Unified model
Posted by Reuven Lax <re...@google.com>.
I'm simply saying that timestamp ordering is insufficient for state
machines. I wasn't proposing Kafka as a solution - that was simply an
example of how people solve this problem in other scenarios.
BTW another example of ordering: Imagine today that you have a triggered
Sum aggregation writing out to a key-value sink. In theory we provide no
ordering, so the sink might write the triggered sums in the wrong order,
ending up with an incorrect value in the sink. In this case you probably
want values ordered by trigger pane index.
Reuven
On Thu, May 23, 2019 at 8:59 AM Jan Lukavský <je...@seznam.cz> wrote:
> Hi Reuven,
> I share the view point of Robert. I think the isuue you refer to is not in
> reality related to timestamps, but to the fact, that ordering of events in
> time is observer dependent (either caused by relativity, or time skew,
> essentially this has the same consequences). And the resolution in fact
> isn't Kafka, but generally an authoritative observer, that tells you "I saw
> the events in this order". And you either have one (and have the outcome of
> his observation persisted in the data - e.g. as offset in Kafka partition),
> then you should be able to use it (maybe that suggests afterall that
> sorting by some user supplied UDF might make sense), or do not have it, and
> then any interpretation of the data seems to be equally valid. Although
> determinism is fine, of course.
> Jan
> ---------- Původní e-mail ----------
> Od: Reuven Lax <re...@google.com>
> Komu: dev <de...@beam.apache.org>
> Datum: 23. 5. 2019 17:39:12
> Předmět: Re: Definition of Unified model
>
> So an example would be elements of type "startUserSession" and
> "endUserSession" (website sessions, not Beam sessions). Logically you may
> need to process them in the correct order if you have any sort of
> state-machine logic. However timestamp ordering is never guaranteed to
> match the logical ordering. Not only might you have several elements with
> the same timestamp, but in reality time skew across backend servers can
> cause the events to have timestamps in reverse order of the actual
> causality order.
>
> People do solve this problem today though. Publish the events to Kafka,
> making sure that events for the same user end up in the same Kafka
> partition. This ensures that the events appear in the Kafka partitions in
> causality order, even if the timestamp order doesn't match. The your Kafka
> subscriber simply process the elements in each partition in order.
>
> I think the ability to impose FIFO causality ordering is what's needed for
> any state-machine work. Timestamp ordering has advantages (though often I
> think the advantage is in state), but does not solve this problem.
>
> Reuven
>
> On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <ro...@google.com>
> wrote:
>
> Good point.
>
> The "implementation-specific" way I would do this is
> window-by-instant, followed by a DoFn that gets all the elements with
> the same timestamp and sorts/acts accordingly, but this counts on the
> runner producing windows in timestamp order (likely?) and also the
> subsequent DoFn getting them in this order (also likely, due to
> fusion).
>
> One could make the argument that, though it does not provide
> deterministic behavior, getting elements of the same timestamp in
> different orders should produce equally valid interpretations of the
> data. (After all, due to relatively, timestamps are not technically
> well ordered across space.) I can see how data-dependent tiebreakers
> could be useful, or promises of preservation of order between
> operations.
>
> - Robert
>
> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com> wrote:
> >
> > So Jan's example of state machines is quite a valid use case for
> ordering. However in my experience, timestamp ordering is insufficient for
> state machines. Elements that cause state transitions might come in with
> the exact same timestamp, yet still have a necessary ordering. Especially
> given Beam's decision to have milliseconds timestamps this is possible, but
> even at microsecond or nanosecond precision this can happen at scale. To
> handle state machines you usually need some sort of FIFO ordering along
> with an ordered sources, such as Kafka, not timestamp ordering.
> >
> > Reuven
> >
> > On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz> wrote:
> >>
> >> Hi all,
> >>
> >> thanks everyone for this discussion. I think I have gathered enough
> >> feedback to be able to put down a proposition for changes, which I will
> >> do and send to this list for further discussion. There are still doubts
> >> remaining the non-determinism and it's relation to outputs stability vs.
> >> latency. But I will try to clarify all this in the design document.
> >>
> >> Thanks,
> >>
> >> Jan
> >>
> >> On 5/22/19 3:49 PM, Maximilian Michels wrote:
> >> >> Someone from Flink might correct me if I'm wrong, but that's my
> >> >> current understanding.
> >> >
> >> > In essence your description of how exactly-once works in Flink is
> >> > correct. The general assumption in Flink is that pipelines must be
> >> > deterministic and thus produce idempotent writes in the case of
> >> > failures. However, that doesn't mean Beam sinks can't guarantee a bit
> >> > more with what Flink has to offer.
> >> >
> >> > Luke already mentioned the design discussions for @RequiresStableInput
> >> > which ensures idempotent writes for non-deterministic pipelines. This
> >> > is not part of the model but an optional Beam feature.
> >> >
> >> > We recently implemented support for @RequiresStableInput in the Flink
> >> > Runner. Reuven mentioned the Flink checkpoint confirmation, which
> >> > allows us to buffer (and checkpoint) processed data and only emit it
> >> > once a Flink checkpoint has completed.
> >> >
> >> > Cheers,
> >> > Max
> >> >
> >> > On 21.05.19 16:49, Jan Lukavský wrote:
> >> >> Hi,
> >> >>
> >> >> > Actually, I think it is a larger (open) question whether exactly
> >> >> once is guaranteed by the model or whether runners are allowed to
> >> >> relax that. I would think, however, that sources correctly
> >> >> implemented should be idempotent when run atop an exactly once
> >> >> infrastructure such as Flink of Dataflow.
> >> >>
> >> >> I would assume, that the model basically inherits guarantees of
> >> >> underlying infrastructure. Because Flink does not work as you
> >> >> described (atomic commit of inputs, state and outputs), but rather a
> >> >> checkpoint mark is flowing through the DAG much like watermark and on
> >> >> failures operators are restored and data reprocessed, it (IMHO)
> >> >> implies, that you have exactly once everywhere in the DAG *but*
> >> >> sinks. That is because sinks cannot be restored to previous state,
> >> >> instead sinks are supposed to be idempotent in order for the exactly
> >> >> once to really work (or at least be able to commit outputs on
> >> >> checkpoint in sink). That implies that if you don't have sink that is
> >> >> able to commit outputs atomically on checkpoint, the pipeline
> >> >> execution should be deterministic upon retries, otherwise shadow
> >> >> writes from failed paths of the pipeline might appear.
> >> >>
> >> >> Someone from Flink might correct me if I'm wrong, but that's my
> >> >> current understanding.
> >> >>
> >> >> > Sounds like we should make this clearer.
> >> >>
> >> >> I meant that you are right that we must not in any thoughts we are
> >> >> having forget that streams are by definition out-of-order. That is
> >> >> property that we cannot change. But - that doesn't limit us from
> >> >> creating operator that presents the data to UDF as if the stream was
> >> >> ideally sorted. It can do that by introducing latency, of course.
> >> >>
> >> >> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
> >> >>> Reza: One could provide something like this as a utility class, but
> >> >>> one downside is that it is not scale invariant. It requires a tuning
> >> >>> parameter that, if to small, won't mitigate the problem, but if to
> >> >>> big, greatly increases latency. (Possibly one could define a dynamic
> >> >>> session-like window to solve this though...) It also might be harder
> >> >>> for runners that *can* cheaply present stuff in timestamp order to
> >> >>> optimize. (That and, in practice, our annotation-style process
> methods
> >> >>> don't lend themselves to easy composition.) I think it could work in
> >> >>> specific cases though.
> >> >>>
> >> >>> More inline below.
> >> >>>
> >> >>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz>
> wrote:
> >> >>>> Hi Robert,
> >> >>>>
> >> >>>> > Beam has an exactly-once model. If the data was consumed, state
> >> >>>> mutated, and outputs written downstream (these three are committed
> >> >>>> together atomically) it will not be replayed. That does not, of
> >> >>>> course,
> >> >>>> solve the non-determanism due to ordering (including the fact that
> two
> >> >>>> operations reading the same PCollection may view different
> ordering).
> >> >>>>
> >> >>>> I think what you describe is a property of a runner, not of the
> model,
> >> >>>> right? I think if I run my pipeline on Flink I will not get this
> >> >>>> atomicity, because although Flink uses also exactly-once model if
> >> >>>> might
> >> >>>> write outputs multiple times.
> >> >>> Actually, I think it is a larger (open) question whether exactly
> once
> >> >>> is guaranteed by the model or whether runners are allowed to relax
> >> >>> that. I would think, however, that sources correctly implemented
> >> >>> should be idempotent when run atop an exactly once infrastructure
> such
> >> >>> as Flink of Dataflow.
> >> >>>
> >> >>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
> >> >>>> received
> >> >>>> in a specific order? In the current model, it is not. Being able to
> >> >>>> read, handle, and produced out-of-order data, including late data,
> >> >>>> is a
> >> >>>> pretty fundamental property of distributed systems.
> >> >>>>
> >> >>>> Yes, absolutely. The argument here is not that Stateful ParDo
> should
> >> >>>> presume to receive elements in any order, but to _present_ it as
> >> >>>> such to
> >> >>>> the user @ProcessElement function.
> >> >>> Sounds like we should make this clearer.
> >> >>>
> >> >>>> > 2) Given that some operations are easier (or possibly only
> >> >>>> possible)
> >> >>>> to write when operating on ordered data, and that different runners
> >> >>>> may
> >> >>>> have (significantly) cheaper ways to provide this ordering than
> can be
> >> >>>> done by the user themselves, should we elevate this to a property
> of
> >> >>>> (Stateful?)DoFns that the runner can provide? I think a compelling
> >> >>>> argument can be made here that we should.
> >> >>>>
> >> >>>> +1
> >> >>>>
> >> >>>> Jan
> >> >>>>
> >> >>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
> >> >>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >> >>>>>> > I don't see batch vs. streaming as part of the model. One
> >> >>>>>> can have
> >> >>>>>> microbatch, or even a runner that alternates between different
> >> >>>>>> modes.
> >> >>>>>>
> >> >>>>>> Although I understand motivation of this statement, this project
> >> >>>>>> name is
> >> >>>>>> "Apache Beam: An advanced unified programming model". What does
> the
> >> >>>>>> model unify, if "streaming vs. batch" is not part of the model?
> >> >>>>> What I mean is that streaming vs. batch is no longer part of the
> >> >>>>> model
> >> >>>>> (or ideally API), but pushed down to be a concern of the runner
> >> >>>>> (executor) of the pipeline.
> >> >>>>>
> >> >>>>>
> >> >>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz>
> >> >>>>> wrote:
> >> >>>>>> Hi Kenn,
> >> >>>>>>
> >> >>>>>> OK, so if we introduce annotation, we can have stateful ParDo
> >> >>>>>> with sorting, that would perfectly resolve my issues. I still
> >> >>>>>> have some doubts, though. Let me explain. The current behavior of
> >> >>>>>> stateful ParDo has the following properties:
> >> >>>>>>
> >> >>>>>> a) might fail in batch, although runs fine in streaming (that
> >> >>>>>> is due to the buffering, and unbounded lateness in batch, which
> >> >>>>>> was discussed back and forth in this thread)
> >> >>>>>>
> >> >>>>>> b) might be non deterministic (this is because the elements
> >> >>>>>> arrive at somewhat random order, and even if you do the operation
> >> >>>>>> "assign unique ID to elements" this might produce different
> >> >>>>>> results when run multiple times)
> >> >>>>> PCollections are *explicitly* unordered. Any operations that
> >> >>>>> assume or
> >> >>>>> depend on a specific ordering for correctness (or determinism)
> must
> >> >>>>> provide that ordering themselves (i.e. tolerate "arbitrary
> shuffling
> >> >>>>> of inputs"). As you point out, that may be very expensive if you
> have
> >> >>>>> very hot keys with very large (unbounded) timestamp skew.
> >> >>>>>
> >> >>>>> StatefulDoFns are low-level operations that should be used with
> care;
> >> >>>>> the simpler windowing model gives determinism in the face of
> >> >>>>> unordered
> >> >>>>> data (though late data and non-end-of-window triggering introduces
> >> >>>>> some of the non-determanism back in).
> >> >>>>>
> >> >>>>>> What worries me most is the property b), because it seems to me
> >> >>>>>> to have serious consequences - not only that if you run twice
> >> >>>>>> batch pipeline you would get different results, but even on
> >> >>>>>> streaming, when pipeline fails and gets restarted from
> >> >>>>>> checkpoint, produced output might differ from the previous run
> >> >>>>>> and data from the first run might have already been persisted
> >> >>>>>> into sink. That would create somewhat messy outputs.
> >> >>>>> Beam has an exactly-once model. If the data was consumed, state
> >> >>>>> mutated, and outputs written downstream (these three are committed
> >> >>>>> together atomically) it will not be replayed. That does not, of
> >> >>>>> course, solve the non-determanism due to ordering (including the
> fact
> >> >>>>> that two operations reading the same PCollection may view
> different
> >> >>>>> ordering).
> >> >>>>>
> >> >>>>>> These two properties makes me think that the current
> >> >>>>>> implementation is more of a _special case_ than the general one.
> >> >>>>>> The general one would be that your state doesn't have the
> >> >>>>>> properties to be able to tolerate buffering problems and/or
> >> >>>>>> non-determinism. Which is the case where you need sorting in both
> >> >>>>>> streaming and batch to be part of the model.
> >> >>>>>>
> >> >>>>>> Let me point out one more analogy - that is merging vs.
> >> >>>>>> non-merging windows. The general case (merging windows) implies
> >> >>>>>> sorting by timestamp in both batch case (explicit) and streaming
> >> >>>>>> (buffering). The special case (non-merging windows) doesn't rely
> >> >>>>>> on any timestamp ordering, so the sorting and buffering can be
> >> >>>>>> dropped. The underlying root cause of this is the same for both
> >> >>>>>> stateful ParDo and windowing (essentially, assigning window
> >> >>>>>> labels is a stateful operation when windowing function is
> merging).
> >> >>>>>>
> >> >>>>>> The reason for the current behavior of stateful ParDo seems to be
> >> >>>>>> performance, but is it right to abandon correctness in favor of
> >> >>>>>> performance? Wouldn't it be more consistent to have the default
> >> >>>>>> behavior prefer correctness and when you have the specific
> >> >>>>>> conditions of state function having special properties, then you
> >> >>>>>> can annotate your DoFn (with something like
> >> >>>>>> @TimeOrderingAgnostic), which would yield a better performance in
> >> >>>>>> that case?
> >> >>>>> There are two separable questions here.
> >> >>>>>
> >> >>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are
> received
> >> >>>>> in a specific order? In the current model, it is not. Being able
> to
> >> >>>>> read, handle, and produced out-of-order data, including late
> data, is
> >> >>>>> a pretty fundamental property of distributed systems.
> >> >>>>>
> >> >>>>> 2) Given that some operations are easier (or possibly only
> possible)
> >> >>>>> to write when operating on ordered data, and that different
> runners
> >> >>>>> may have (significantly) cheaper ways to provide this ordering
> than
> >> >>>>> can be done by the user themselves, should we elevate this to a
> >> >>>>> property of (Stateful?)DoFns that the runner can provide? I think
> a
> >> >>>>> compelling argument can be made here that we should.
> >> >>>>>
> >> >>>>> - Robert
> >> >>>>>
> >> >>>>>
> >> >>>>>
> >> >>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
> >> >>>>>>
> >> >>>>>> Thanks for the nice small example of a calculation that depends
> >> >>>>>> on order. You are right that many state machines have this
> >> >>>>>> property. I agree w/ you and Luke that it is convenient for batch
> >> >>>>>> processing to sort by event timestamp before running a stateful
> >> >>>>>> ParDo. In streaming you could also implement "sort by event
> >> >>>>>> timestamp" by buffering until you know all earlier data will be
> >> >>>>>> dropped - a slack buffer up to allowed lateness.
> >> >>>>>>
> >> >>>>>> I do not think that it is OK to sort in batch and not in
> >> >>>>>> streaming. Many state machines diverge very rapidly when things
> >> >>>>>> are out of order. So each runner if they see the
> >> >>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
> >> >>>>>> sorted data (by some mix of buffering and dropping), or to reject
> >> >>>>>> the pipeline as unsupported.
> >> >>>>>>
> >> >>>>>> And also want to say that this is not the default case - many
> >> >>>>>> uses of state & timers in ParDo yield different results at the
> >> >>>>>> element level, but the results are equivalent at in the big
> >> >>>>>> picture. Such as the example of "assign a unique sequence number
> >> >>>>>> to each element" or "group into batches" it doesn't matter
> >> >>>>>> exactly what the result is, only that it meets the spec. And
> >> >>>>>> other cases like user funnels are monotonic enough that you also
> >> >>>>>> don't actually need sorting.
> >> >>>>>>
> >> >>>>>> Kenn
> >> >>>>>>
> >> >>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz>
> >> >>>>>> wrote:
> >> >>>>>>> Yes, the problem will arise probably mostly when you have not
> >> >>>>>>> well distributed keys (or too few keys). I'm really not sure if
> >> >>>>>>> a pure GBK with a trigger can solve this - it might help to have
> >> >>>>>>> data driven trigger. There would still be some doubts, though.
> >> >>>>>>> The main question is still here - people say, that sorting by
> >> >>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
> >> >>>>>>> I don't really see why - the sorting is very probably already
> >> >>>>>>> there. And if not (hash grouping instead of sorted grouping),
> >> >>>>>>> then the sorting would affect only user defined StatefulParDos.
> >> >>>>>>>
> >> >>>>>>> This would suggest that the best way out of this would be really
> >> >>>>>>> to add annotation, so that the author of the pipeline can
> decide.
> >> >>>>>>>
> >> >>>>>>> If that would be acceptable I think I can try to prepare some
> >> >>>>>>> basic functionality, but I'm not sure, if I would be able to
> >> >>>>>>> cover all runners / sdks.
> >> >>>>>>>
> >> >>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
> >> >>>>>>>
> >> >>>>>>> It is read all per key and window and not just read all (this
> >> >>>>>>> still won't scale with hot keys in the global window). The GBK
> >> >>>>>>> preceding the StatefulParDo will guarantee that you are
> >> >>>>>>> processing all the values for a specific key and window at any
> >> >>>>>>> given time. Is there a specific window/trigger that is missing
> >> >>>>>>> that you feel would remove the need for you to use
> StatefulParDo?
> >> >>>>>>>
> >> >>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
> >> >>>>>>> wrote:
> >> >>>>>>>> Hi Lukasz,
> >> >>>>>>>>
> >> >>>>>>>>> Today, if you must have a strict order, you must guarantee
> >> >>>>>>>>> that your StatefulParDo implements the necessary "buffering &
> >> >>>>>>>>> sorting" into state.
> >> >>>>>>>> Yes, no problem with that. But this whole discussion started,
> >> >>>>>>>> because *this doesn't work on batch*. You simply cannot first
> >> >>>>>>>> read everything from distributed storage and then buffer it all
> >> >>>>>>>> into memory, just to read it again, but sorted. That will not
> >> >>>>>>>> work. And even if it would, it would be a terrible waste of
> >> >>>>>>>> resources.
> >> >>>>>>>>
> >> >>>>>>>> Jan
> >> >>>>>>>>
> >> >>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
> >> >>>>>>>>
> >> >>>>>>>>
> >> >>>>>>>>
> >> >>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
> >> >>>>>>>> wrote:
> >> >>>>>>>>> This discussion brings many really interesting questions for
> >> >>>>>>>>> me. :-)
> >> >>>>>>>>>
> >> >>>>>>>>> > I don't see batch vs. streaming as part of the model. One
> >> >>>>>>>>> can have
> >> >>>>>>>>> microbatch, or even a runner that alternates between different
> >> >>>>>>>>> modes.
> >> >>>>>>>>>
> >> >>>>>>>>> Although I understand motivation of this statement, this
> >> >>>>>>>>> project name is
> >> >>>>>>>>> "Apache Beam: An advanced unified programming model". What
> >> >>>>>>>>> does the
> >> >>>>>>>>> model unify, if "streaming vs. batch" is not part of the
> model?
> >> >>>>>>>>>
> >> >>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
> >> >>>>>>>>> are
> >> >>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
> >> >>>>>>>>> All these
> >> >>>>>>>>> define several runtime parameters, which in turn define how
> >> >>>>>>>>> well/badly
> >> >>>>>>>>> will the pipeline perform and how many resources might be
> >> >>>>>>>>> needed. From
> >> >>>>>>>>> my point of view, pure streaming should be the most resource
> >> >>>>>>>>> demanding
> >> >>>>>>>>> (if not, why bother with batch? why not run everything in
> >> >>>>>>>>> streaming
> >> >>>>>>>>> only? what will there remain to "unify"?).
> >> >>>>>>>>>
> >> >>>>>>>>> > Fortunately, for batch, only the state for a single key
> >> >>>>>>>>> needs to be
> >> >>>>>>>>> preserved at a time, rather than the state for all keys across
> >> >>>>>>>>> the range
> >> >>>>>>>>> of skew. Of course if you have few or hot keys, one can still
> >> >>>>>>>>> have
> >> >>>>>>>>> issues (and this is not specific to StatefulDoFns).
> >> >>>>>>>>>
> >> >>>>>>>>> Yes, but here is still the presumption that my stateful DoFn
> can
> >> >>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
> >> >>>>>>>>> case in
> >> >>>>>>>>> more detail.
> >> >>>>>>>>>
> >> >>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
> >> >>>>>>>>> some key for
> >> >>>>>>>>> each element, which is irrelevant for the demonstration). Your
> >> >>>>>>>>> task is
> >> >>>>>>>>> to calculate in running global window the actual number of
> >> >>>>>>>>> changes
> >> >>>>>>>>> between state 0 and state 1 and vice versa. When the state
> >> >>>>>>>>> doesn't
> >> >>>>>>>>> change, you don't calculate anything. If input (for given key)
> >> >>>>>>>>> would be
> >> >>>>>>>>> (tN denotes timestamp N):
> >> >>>>>>>>>
> >> >>>>>>>>> t1: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t2: 0
> >> >>>>>>>>>
> >> >>>>>>>>> t3: 0
> >> >>>>>>>>>
> >> >>>>>>>>> t4: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t5: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t6: 0
> >> >>>>>>>>>
> >> >>>>>>>>> then the output should yield (supposing that default state is
> >> >>>>>>>>> zero):
> >> >>>>>>>>>
> >> >>>>>>>>> t1: (one: 1, zero: 0)
> >> >>>>>>>>>
> >> >>>>>>>>> t2: (one: 1, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t3: (one: 1, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t4: (one: 2, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t5: (one: 2, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t6: (one: 2, zero: 2)
> >> >>>>>>>>>
> >> >>>>>>>>> How would you implement this in current Beam semantics?
> >> >>>>>>>> I think your saying here that I know that my input is ordered
> >> >>>>>>>> in a specific way and since I assume the order when writing my
> >> >>>>>>>> pipeline I can perform this optimization. But there is nothing
> >> >>>>>>>> preventing a runner from noticing that your processing in the
> >> >>>>>>>> global window with a specific type of trigger and re-ordering
> >> >>>>>>>> your inputs/processing to get better performance (since you
> >> >>>>>>>> can't use an AfterWatermark trigger for your pipeline in
> >> >>>>>>>> streaming for the GlobalWindow).
> >> >>>>>>>>
> >> >>>>>>>> Today, if you must have a strict order, you must guarantee that
> >> >>>>>>>> your StatefulParDo implements the necessary "buffering &
> >> >>>>>>>> sorting" into state. I can see why you would want an annotation
> >> >>>>>>>> that says I must have timestamp ordered elements, since it
> >> >>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
> >> >>>>>>>> is a low-level function, it really is the "here you go and do
> >> >>>>>>>> whatever you need to but here be dragons" function while
> >> >>>>>>>> windowing and triggering is meant to keep many people from
> >> >>>>>>>> writing StatefulParDo in the first place.
> >> >>>>>>>>
> >> >>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
> >> >>>>>>>>> are likely to
> >> >>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
> >> >>>>>>>>> falls
> >> >>>>>>>>> behind in streaming mode as well.
> >> >>>>>>>>>
> >> >>>>>>>>> But the worst case is defined by input of size (available
> >> >>>>>>>>> resources +
> >> >>>>>>>>> single byte) -> pipeline fail. Although it could have
> >> >>>>>>>>> finished, given
> >> >>>>>>>>> the right conditions.
> >> >>>>>>>>>
> >> >>>>>>>>> > This might be reasonable, implemented by default by
> >> >>>>>>>>> buffering
> >> >>>>>>>>> everything and releasing elements as the watermark (+lateness)
> >> >>>>>>>>> advances,
> >> >>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
> >> >>>>>>>>> reason
> >> >>>>>>>>> about) code.
> >> >>>>>>>>>
> >> >>>>>>>>> Sure, the pipeline will be less efficient, because it would
> >> >>>>>>>>> have to
> >> >>>>>>>>> buffer and sort the inputs. But at least it will produce
> >> >>>>>>>>> correct results
> >> >>>>>>>>> in cases where updates to state are order-sensitive.
> >> >>>>>>>>>
> >> >>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
> >> >>>>>>>>> (key, values):
> >> >>>>>>>>> [(key, value) for value in values])?
> >> >>>>>>>>>
> >> >>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
> >> >>>>>>>>> The trigger
> >> >>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
> >> >>>>>>>>> element
> >> >>>>>>>>> with lowest timestamp in the buffer. Although this could be
> >> >>>>>>>>> somehow
> >> >>>>>>>>> emulated by fixed trigger each X millis.
> >> >>>>>>>>>
> >> >>>>>>>>> > Or is the underlying desire just to be able to hint to
> >> >>>>>>>>> the runner
> >> >>>>>>>>> that the code may perform better (e.g. require less resources)
> >> >>>>>>>>> as skew
> >> >>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
> >> >>>>>>>>>
> >> >>>>>>>>> No, the sorting would have to be done in streaming case as
> >> >>>>>>>>> well. That is
> >> >>>>>>>>> an imperative of the unified model. I think it is possible to
> >> >>>>>>>>> sort by
> >> >>>>>>>>> timestamp only in batch case (and do it for *all* batch
> >> >>>>>>>>> stateful pardos
> >> >>>>>>>>> without annotation), or introduce annotation, but then make
> >> >>>>>>>>> the same
> >> >>>>>>>>> guarantees for streaming case as well.
> >> >>>>>>>>>
> >> >>>>>>>>> Jan
> >> >>>>>>>>>
> >> >>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
> >> >>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
> >> >>>>>>>>>> <je...@seznam.cz> wrote:
> >> >>>>>>>>>>> Hi Robert,
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
> >> >>>>>>>>>>> guarantees
> >> >>>>>>>>>>> of ordering are given in either mode, there is *implicit*
> >> >>>>>>>>>>> ordering in
> >> >>>>>>>>>>> streaming case that is due to nature of the processing - the
> >> >>>>>>>>>>> difference
> >> >>>>>>>>>>> between watermark and timestamp of elements flowing through
> >> >>>>>>>>>>> the pipeline
> >> >>>>>>>>>>> are generally low (too high difference leads to the
> >> >>>>>>>>>>> overbuffering
> >> >>>>>>>>>>> problem), but there is no such bound on batch.
> >> >>>>>>>>>> Fortunately, for batch, only the state for a single key needs
> >> >>>>>>>>>> to be
> >> >>>>>>>>>> preserved at a time, rather than the state for all keys
> >> >>>>>>>>>> across the
> >> >>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
> >> >>>>>>>>>> still
> >> >>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
> >> >>>>>>>>>>
> >> >>>>>>>>>>> As a result, I see a few possible solutions:
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> - the best and most natural seems to be extension of
> >> >>>>>>>>>>> the model, so
> >> >>>>>>>>>>> that it defines batch as not only "streaming pipeline
> >> >>>>>>>>>>> executed in batch
> >> >>>>>>>>>>> fashion", but "pipeline with at least as good runtime
> >> >>>>>>>>>>> characteristics as
> >> >>>>>>>>>>> in streaming case, executed in batch fashion", I really
> >> >>>>>>>>>>> don't think that
> >> >>>>>>>>>>> there are any conflicts with the current model, or that this
> >> >>>>>>>>>>> could
> >> >>>>>>>>>>> affect performance, because the required sorting (as
> pointed by
> >> >>>>>>>>>>> Aljoscha) is very probably already done during translation
> >> >>>>>>>>>>> of stateful
> >> >>>>>>>>>>> pardos. Also note that this definition only affects user
> >> >>>>>>>>>>> defined
> >> >>>>>>>>>>> stateful pardos
> >> >>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
> >> >>>>>>>>>> have
> >> >>>>>>>>>> microbatch, or even a runner that alternates between
> >> >>>>>>>>>> different modes.
> >> >>>>>>>>>> The model describes what the valid outputs are given a
> >> >>>>>>>>>> (sometimes
> >> >>>>>>>>>> partial) set of inputs. It becomes really hard to define
> >> >>>>>>>>>> things like
> >> >>>>>>>>>> "as good runtime characteristics." Once you allow any
> >> >>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
> >> >>>>>>>>>> (and
> >> >>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
> >> >>>>>>>>>> out-of-orderedness.
> >> >>>>>>>>>>
> >> >>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
> >> >>>>>>>>>> likely to
> >> >>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
> >> >>>>>>>>>> falls
> >> >>>>>>>>>> behind in streaming mode as well.
> >> >>>>>>>>>>
> >> >>>>>>>>>>> - another option would be to introduce annotation for
> >> >>>>>>>>>>> DoFns (e.g.
> >> >>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
> >> >>>>>>>>>>> the sorting
> >> >>>>>>>>>>> in batch case - but - this extension would have to ensure
> >> >>>>>>>>>>> the sorting in
> >> >>>>>>>>>>> streaming mode also - it would require definition of allowed
> >> >>>>>>>>>>> lateness,
> >> >>>>>>>>>>> and triggger (essentially similar to window)
> >> >>>>>>>>>> This might be reasonable, implemented by default by buffering
> >> >>>>>>>>>> everything and releasing elements as the watermark
> (+lateness)
> >> >>>>>>>>>> advances, but would likely lead to inefficient (though
> >> >>>>>>>>>> *maybe* easier
> >> >>>>>>>>>> to reason about) code. Not sure about the semantics of
> >> >>>>>>>>>> triggering
> >> >>>>>>>>>> here, especially data-driven triggers. Would it be roughly
> >> >>>>>>>>>> equivalent
> >> >>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
> >> >>>>>>>>>> value in
> >> >>>>>>>>>> values])?
> >> >>>>>>>>>>
> >> >>>>>>>>>> Or is the underlying desire just to be able to hint to the
> >> >>>>>>>>>> runner that
> >> >>>>>>>>>> the code may perform better (e.g. require less resources) as
> >> >>>>>>>>>> skew is
> >> >>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
> >> >>>>>>>>>>
> >> >>>>>>>>>>> - last option would be to introduce these "higher order
> >> >>>>>>>>>>> guarantees" in
> >> >>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
> >> >>>>>>>>>>> worst
> >> >>>>>>>>>>> option to me
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> I see the first two options quite equally good, although the
> >> >>>>>>>>>>> letter one
> >> >>>>>>>>>>> is probably more time consuming to implement. But it would
> >> >>>>>>>>>>> bring
> >> >>>>>>>>>>> additional feature to streaming case as well.
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> Thanks for any thoughts.
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> Jan
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
> >> >>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
> >> >>>>>>>>>>>> <je...@seznam.cz> wrote:
> >> >>>>>>>>>>>>> Hi Reuven,
> >> >>>>>>>>>>>>>
> >> >>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
> >> >>>>>>>>>>>>>> runners.
> >> >>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
> >> >>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
> >> >>>>>>>>>>>>> of elements. That basically (practically) can work only
> >> >>>>>>>>>>>>> for cases, where the order of input elements doesn't
> >> >>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
> >> >>>>>>>>>>>>> time you have a state machine involved, then the ordering
> >> >>>>>>>>>>>>> of elements would matter.
> >> >>>>>>>>>>>> No guarantees on order are provided in *either* streaming
> >> >>>>>>>>>>>> or batch
> >> >>>>>>>>>>>> mode by the model. However, it is the case that in order to
> >> >>>>>>>>>>>> make
> >> >>>>>>>>>>>> forward progress most streaming runners attempt to limit
> >> >>>>>>>>>>>> the amount of
> >> >>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
> >> >>>>>>>>>>>> processing
> >> >>>>>>>>>>>> time) to make forward progress, which in turn could help
> >> >>>>>>>>>>>> cap the
> >> >>>>>>>>>>>> amount of state that must be held concurrently, whereas a
> >> >>>>>>>>>>>> batch runner
> >> >>>>>>>>>>>> may not allow any state to be safely discarded until the
> whole
> >> >>>>>>>>>>>> timeline from infinite past to infinite future has been
> >> >>>>>>>>>>>> observed.
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
> >> >>>>>>>>>>>> batch" in batch mode.
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
> >> >>>>>>>>>>>> <mx...@apache.org> wrote:
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>>>> batch semantics and streaming semantics differs only
> >> >>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
> >> >>>>>>>>>>>>>> batch and cannot on stream
> >> >>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
> >> >>>>>>>>>>>>> trigger. You
> >> >>>>>>>>>>>>> could define additional triggers that do early firings.
> >> >>>>>>>>>>>>> And you could
> >> >>>>>>>>>>>>> even trigger the global window by advancing the watermark
> >> >>>>>>>>>>>>> to +inf.
> >> >>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
> >> >>>>>>>>>>>> default
> >> >>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
> >> >>>>>>>>>>>> is more
> >> >>>>>>>>>>>> likely to be user error than an actual desire to have no
> >> >>>>>>>>>>>> output until
> >> >>>>>>>>>>>> drain. But it's semantically valid in the model.
>
>
Re: Definition of Unified model
Posted by Jan Lukavský <je...@seznam.cz>.
Hi Reuven,
I share the view point of Robert. I think the isuue you refer to is not in
reality related to timestamps, but to the fact, that ordering of events in
time is observer dependent (either caused by relativity, or time skew,
essentially this has the same consequences). And the resolution in fact isn'
t Kafka, but generally an authoritative observer, that tells you "I saw the
events in this order". And you either have one (and have the outcome of his
observation persisted in the data - e.g. as offset in Kafka partition), then
you should be able to use it (maybe that suggests afterall that sorting by
some user supplied UDF might make sense), or do not have it, and then any
interpretation of the data seems to be equally valid. Although determinism
is fine, of course.
Jan
---------- Původní e-mail ----------
Od: Reuven Lax <re...@google.com>
Komu: dev <de...@beam.apache.org>
Datum: 23. 5. 2019 17:39:12
Předmět: Re: Definition of Unified model
"
So an example would be elements of type "startUserSession" and
"endUserSession" (website sessions, not Beam sessions). Logically you may
need to process them in the correct order if you have any sort of state-
machine logic. However timestamp ordering is never guaranteed to match the
logical ordering. Not only might you have several elements with the same
timestamp, but in reality time skew across backend servers can cause the
events to have timestamps in reverse order of the actual causality order.
People do solve this problem today though. Publish the events to Kafka,
making sure that events for the same user end up in the same Kafka
partition. This ensures that the events appear in the Kafka partitions in
causality order, even if the timestamp order doesn't match. The your Kafka
subscriber simply process the elements in each partition in order.
I think the ability to impose FIFO causality ordering is what's needed for
any state-machine work. Timestamp ordering has advantages (though often I
think the advantage is in state), but does not solve this problem.
Reuven
On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <robertwb@google.com
(mailto:robertwb@google.com)> wrote:
"Good point.
The "implementation-specific" way I would do this is
window-by-instant, followed by a DoFn that gets all the elements with
the same timestamp and sorts/acts accordingly, but this counts on the
runner producing windows in timestamp order (likely?) and also the
subsequent DoFn getting them in this order (also likely, due to
fusion).
One could make the argument that, though it does not provide
deterministic behavior, getting elements of the same timestamp in
different orders should produce equally valid interpretations of the
data. (After all, due to relatively, timestamps are not technically
well ordered across space.) I can see how data-dependent tiebreakers
could be useful, or promises of preservation of order between
operations.
- Robert
On Thu, May 23, 2019 at 4:18 PM Reuven Lax <relax@google.com
(mailto:relax@google.com)> wrote:
>
> So Jan's example of state machines is quite a valid use case for ordering.
However in my experience, timestamp ordering is insufficient for state
machines. Elements that cause state transitions might come in with the exact
same timestamp, yet still have a necessary ordering. Especially given Beam's
decision to have milliseconds timestamps this is possible, but even at
microsecond or nanosecond precision this can happen at scale. To handle
state machines you usually need some sort of FIFO ordering along with an
ordered sources, such as Kafka, not timestamp ordering.
>
> Reuven
>
> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)> wrote:
>>
>> Hi all,
>>
>> thanks everyone for this discussion. I think I have gathered enough
>> feedback to be able to put down a proposition for changes, which I will
>> do and send to this list for further discussion. There are still doubts
>> remaining the non-determinism and it's relation to outputs stability vs.
>> latency. But I will try to clarify all this in the design document.
>>
>> Thanks,
>>
>> Jan
>>
>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>> >> Someone from Flink might correct me if I'm wrong, but that's my
>> >> current understanding.
>> >
>> > In essence your description of how exactly-once works in Flink is
>> > correct. The general assumption in Flink is that pipelines must be
>> > deterministic and thus produce idempotent writes in the case of
>> > failures. However, that doesn't mean Beam sinks can't guarantee a bit
>> > more with what Flink has to offer.
>> >
>> > Luke already mentioned the design discussions for @RequiresStableInput
>> > which ensures idempotent writes for non-deterministic pipelines. This
>> > is not part of the model but an optional Beam feature.
>> >
>> > We recently implemented support for @RequiresStableInput in the Flink
>> > Runner. Reuven mentioned the Flink checkpoint confirmation, which
>> > allows us to buffer (and checkpoint) processed data and only emit it
>> > once a Flink checkpoint has completed.
>> >
>> > Cheers,
>> > Max
>> >
>> > On 21.05.19 16:49, Jan Lukavský wrote:
>> >> Hi,
>> >>
>> >> > Actually, I think it is a larger (open) question whether exactly
>> >> once is guaranteed by the model or whether runners are allowed to
>> >> relax that. I would think, however, that sources correctly
>> >> implemented should be idempotent when run atop an exactly once
>> >> infrastructure such as Flink of Dataflow.
>> >>
>> >> I would assume, that the model basically inherits guarantees of
>> >> underlying infrastructure. Because Flink does not work as you
>> >> described (atomic commit of inputs, state and outputs), but rather a
>> >> checkpoint mark is flowing through the DAG much like watermark and on
>> >> failures operators are restored and data reprocessed, it (IMHO)
>> >> implies, that you have exactly once everywhere in the DAG *but*
>> >> sinks. That is because sinks cannot be restored to previous state,
>> >> instead sinks are supposed to be idempotent in order for the exactly
>> >> once to really work (or at least be able to commit outputs on
>> >> checkpoint in sink). That implies that if you don't have sink that is
>> >> able to commit outputs atomically on checkpoint, the pipeline
>> >> execution should be deterministic upon retries, otherwise shadow
>> >> writes from failed paths of the pipeline might appear.
>> >>
>> >> Someone from Flink might correct me if I'm wrong, but that's my
>> >> current understanding.
>> >>
>> >> > Sounds like we should make this clearer.
>> >>
>> >> I meant that you are right that we must not in any thoughts we are
>> >> having forget that streams are by definition out-of-order. That is
>> >> property that we cannot change. But - that doesn't limit us from
>> >> creating operator that presents the data to UDF as if the stream was
>> >> ideally sorted. It can do that by introducing latency, of course.
>> >>
>> >> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>> >>> Reza: One could provide something like this as a utility class, but
>> >>> one downside is that it is not scale invariant. It requires a tuning
>> >>> parameter that, if to small, won't mitigate the problem, but if to
>> >>> big, greatly increases latency. (Possibly one could define a dynamic
>> >>> session-like window to solve this though...) It also might be harder
>> >>> for runners that *can* cheaply present stuff in timestamp order to
>> >>> optimize. (That and, in practice, our annotation-style process
methods
>> >>> don't lend themselves to easy composition.) I think it could work in
>> >>> specific cases though.
>> >>>
>> >>> More inline below.
>> >>>
>> >>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)> wrote:
>> >>>> Hi Robert,
>> >>>>
>> >>>> > Beam has an exactly-once model. If the data was consumed, state
>> >>>> mutated, and outputs written downstream (these three are committed
>> >>>> together atomically) it will not be replayed. That does not, of
>> >>>> course,
>> >>>> solve the non-determanism due to ordering (including the fact that
two
>> >>>> operations reading the same PCollection may view different
ordering).
>> >>>>
>> >>>> I think what you describe is a property of a runner, not of the
model,
>> >>>> right? I think if I run my pipeline on Flink I will not get this
>> >>>> atomicity, because although Flink uses also exactly-once model if
>> >>>> might
>> >>>> write outputs multiple times.
>> >>> Actually, I think it is a larger (open) question whether exactly once
>> >>> is guaranteed by the model or whether runners are allowed to relax
>> >>> that. I would think, however, that sources correctly implemented
>> >>> should be idempotent when run atop an exactly once infrastructure
such
>> >>> as Flink of Dataflow.
>> >>>
>> >>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
>> >>>> received
>> >>>> in a specific order? In the current model, it is not. Being able to
>> >>>> read, handle, and produced out-of-order data, including late data,
>> >>>> is a
>> >>>> pretty fundamental property of distributed systems.
>> >>>>
>> >>>> Yes, absolutely. The argument here is not that Stateful ParDo should
>> >>>> presume to receive elements in any order, but to _present_ it as
>> >>>> such to
>> >>>> the user @ProcessElement function.
>> >>> Sounds like we should make this clearer.
>> >>>
>> >>>> > 2) Given that some operations are easier (or possibly only
>> >>>> possible)
>> >>>> to write when operating on ordered data, and that different runners
>> >>>> may
>> >>>> have (significantly) cheaper ways to provide this ordering than can
be
>> >>>> done by the user themselves, should we elevate this to a property of
>> >>>> (Stateful?)DoFns that the runner can provide? I think a compelling
>> >>>> argument can be made here that we should.
>> >>>>
>> >>>> +1
>> >>>>
>> >>>> Jan
>> >>>>
>> >>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>> >>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)> wrote:
>> >>>>>> > I don't see batch vs. streaming as part of the model. One
>> >>>>>> can have
>> >>>>>> microbatch, or even a runner that alternates between different
>> >>>>>> modes.
>> >>>>>>
>> >>>>>> Although I understand motivation of this statement, this project
>> >>>>>> name is
>> >>>>>> "Apache Beam: An advanced unified programming model". What does
the
>> >>>>>> model unify, if "streaming vs. batch" is not part of the model?
>> >>>>> What I mean is that streaming vs. batch is no longer part of the
>> >>>>> model
>> >>>>> (or ideally API), but pushed down to be a concern of the runner
>> >>>>> (executor) of the pipeline.
>> >>>>>
>> >>>>>
>> >>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)>
>> >>>>> wrote:
>> >>>>>> Hi Kenn,
>> >>>>>>
>> >>>>>> OK, so if we introduce annotation, we can have stateful ParDo
>> >>>>>> with sorting, that would perfectly resolve my issues. I still
>> >>>>>> have some doubts, though. Let me explain. The current behavior of
>> >>>>>> stateful ParDo has the following properties:
>> >>>>>>
>> >>>>>> a) might fail in batch, although runs fine in streaming (that
>> >>>>>> is due to the buffering, and unbounded lateness in batch, which
>> >>>>>> was discussed back and forth in this thread)
>> >>>>>>
>> >>>>>> b) might be non deterministic (this is because the elements
>> >>>>>> arrive at somewhat random order, and even if you do the operation
>> >>>>>> "assign unique ID to elements" this might produce different
>> >>>>>> results when run multiple times)
>> >>>>> PCollections are *explicitly* unordered. Any operations that
>> >>>>> assume or
>> >>>>> depend on a specific ordering for correctness (or determinism) must
>> >>>>> provide that ordering themselves (i.e. tolerate "arbitrary
shuffling
>> >>>>> of inputs"). As you point out, that may be very expensive if you
have
>> >>>>> very hot keys with very large (unbounded) timestamp skew.
>> >>>>>
>> >>>>> StatefulDoFns are low-level operations that should be used with
care;
>> >>>>> the simpler windowing model gives determinism in the face of
>> >>>>> unordered
>> >>>>> data (though late data and non-end-of-window triggering introduces
>> >>>>> some of the non-determanism back in).
>> >>>>>
>> >>>>>> What worries me most is the property b), because it seems to me
>> >>>>>> to have serious consequences - not only that if you run twice
>> >>>>>> batch pipeline you would get different results, but even on
>> >>>>>> streaming, when pipeline fails and gets restarted from
>> >>>>>> checkpoint, produced output might differ from the previous run
>> >>>>>> and data from the first run might have already been persisted
>> >>>>>> into sink. That would create somewhat messy outputs.
>> >>>>> Beam has an exactly-once model. If the data was consumed, state
>> >>>>> mutated, and outputs written downstream (these three are committed
>> >>>>> together atomically) it will not be replayed. That does not, of
>> >>>>> course, solve the non-determanism due to ordering (including the
fact
>> >>>>> that two operations reading the same PCollection may view different
>> >>>>> ordering).
>> >>>>>
>> >>>>>> These two properties makes me think that the current
>> >>>>>> implementation is more of a _special case_ than the general one.
>> >>>>>> The general one would be that your state doesn't have the
>> >>>>>> properties to be able to tolerate buffering problems and/or
>> >>>>>> non-determinism. Which is the case where you need sorting in both
>> >>>>>> streaming and batch to be part of the model.
>> >>>>>>
>> >>>>>> Let me point out one more analogy - that is merging vs.
>> >>>>>> non-merging windows. The general case (merging windows) implies
>> >>>>>> sorting by timestamp in both batch case (explicit) and streaming
>> >>>>>> (buffering). The special case (non-merging windows) doesn't rely
>> >>>>>> on any timestamp ordering, so the sorting and buffering can be
>> >>>>>> dropped. The underlying root cause of this is the same for both
>> >>>>>> stateful ParDo and windowing (essentially, assigning window
>> >>>>>> labels is a stateful operation when windowing function is
merging).
>> >>>>>>
>> >>>>>> The reason for the current behavior of stateful ParDo seems to be
>> >>>>>> performance, but is it right to abandon correctness in favor of
>> >>>>>> performance? Wouldn't it be more consistent to have the default
>> >>>>>> behavior prefer correctness and when you have the specific
>> >>>>>> conditions of state function having special properties, then you
>> >>>>>> can annotate your DoFn (with something like
>> >>>>>> @TimeOrderingAgnostic), which would yield a better performance in
>> >>>>>> that case?
>> >>>>> There are two separable questions here.
>> >>>>>
>> >>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are
received
>> >>>>> in a specific order? In the current model, it is not. Being able to
>> >>>>> read, handle, and produced out-of-order data, including late data,
is
>> >>>>> a pretty fundamental property of distributed systems.
>> >>>>>
>> >>>>> 2) Given that some operations are easier (or possibly only
possible)
>> >>>>> to write when operating on ordered data, and that different runners
>> >>>>> may have (significantly) cheaper ways to provide this ordering than
>> >>>>> can be done by the user themselves, should we elevate this to a
>> >>>>> property of (Stateful?)DoFns that the runner can provide? I think a
>> >>>>> compelling argument can be made here that we should.
>> >>>>>
>> >>>>> - Robert
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>> >>>>>>
>> >>>>>> Thanks for the nice small example of a calculation that depends
>> >>>>>> on order. You are right that many state machines have this
>> >>>>>> property. I agree w/ you and Luke that it is convenient for batch
>> >>>>>> processing to sort by event timestamp before running a stateful
>> >>>>>> ParDo. In streaming you could also implement "sort by event
>> >>>>>> timestamp" by buffering until you know all earlier data will be
>> >>>>>> dropped - a slack buffer up to allowed lateness.
>> >>>>>>
>> >>>>>> I do not think that it is OK to sort in batch and not in
>> >>>>>> streaming. Many state machines diverge very rapidly when things
>> >>>>>> are out of order. So each runner if they see the
>> >>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
>> >>>>>> sorted data (by some mix of buffering and dropping), or to reject
>> >>>>>> the pipeline as unsupported.
>> >>>>>>
>> >>>>>> And also want to say that this is not the default case - many
>> >>>>>> uses of state & timers in ParDo yield different results at the
>> >>>>>> element level, but the results are equivalent at in the big
>> >>>>>> picture. Such as the example of "assign a unique sequence number
>> >>>>>> to each element" or "group into batches" it doesn't matter
>> >>>>>> exactly what the result is, only that it meets the spec. And
>> >>>>>> other cases like user funnels are monotonic enough that you also
>> >>>>>> don't actually need sorting.
>> >>>>>>
>> >>>>>> Kenn
>> >>>>>>
>> >>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)>
>> >>>>>> wrote:
>> >>>>>>> Yes, the problem will arise probably mostly when you have not
>> >>>>>>> well distributed keys (or too few keys). I'm really not sure if
>> >>>>>>> a pure GBK with a trigger can solve this - it might help to have
>> >>>>>>> data driven trigger. There would still be some doubts, though.
>> >>>>>>> The main question is still here - people say, that sorting by
>> >>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
>> >>>>>>> I don't really see why - the sorting is very probably already
>> >>>>>>> there. And if not (hash grouping instead of sorted grouping),
>> >>>>>>> then the sorting would affect only user defined StatefulParDos.
>> >>>>>>>
>> >>>>>>> This would suggest that the best way out of this would be really
>> >>>>>>> to add annotation, so that the author of the pipeline can decide.
>> >>>>>>>
>> >>>>>>> If that would be acceptable I think I can try to prepare some
>> >>>>>>> basic functionality, but I'm not sure, if I would be able to
>> >>>>>>> cover all runners / sdks.
>> >>>>>>>
>> >>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>> >>>>>>>
>> >>>>>>> It is read all per key and window and not just read all (this
>> >>>>>>> still won't scale with hot keys in the global window). The GBK
>> >>>>>>> preceding the StatefulParDo will guarantee that you are
>> >>>>>>> processing all the values for a specific key and window at any
>> >>>>>>> given time. Is there a specific window/trigger that is missing
>> >>>>>>> that you feel would remove the need for you to use StatefulParDo?
>> >>>>>>>
>> >>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)>
>> >>>>>>> wrote:
>> >>>>>>>> Hi Lukasz,
>> >>>>>>>>
>> >>>>>>>>> Today, if you must have a strict order, you must guarantee
>> >>>>>>>>> that your StatefulParDo implements the necessary "buffering &
>> >>>>>>>>> sorting" into state.
>> >>>>>>>> Yes, no problem with that. But this whole discussion started,
>> >>>>>>>> because *this doesn't work on batch*. You simply cannot first
>> >>>>>>>> read everything from distributed storage and then buffer it all
>> >>>>>>>> into memory, just to read it again, but sorted. That will not
>> >>>>>>>> work. And even if it would, it would be a terrible waste of
>> >>>>>>>> resources.
>> >>>>>>>>
>> >>>>>>>> Jan
>> >>>>>>>>
>> >>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>> >>>>>>>>
>> >>>>>>>>
>> >>>>>>>>
>> >>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je.ik@seznam.cz
(mailto:je.ik@seznam.cz)>
>> >>>>>>>> wrote:
>> >>>>>>>>> This discussion brings many really interesting questions for
>> >>>>>>>>> me. :-)
>> >>>>>>>>>
>> >>>>>>>>> > I don't see batch vs. streaming as part of the model. One
>> >>>>>>>>> can have
>> >>>>>>>>> microbatch, or even a runner that alternates between different
>> >>>>>>>>> modes.
>> >>>>>>>>>
>> >>>>>>>>> Although I understand motivation of this statement, this
>> >>>>>>>>> project name is
>> >>>>>>>>> "Apache Beam: An advanced unified programming model". What
>> >>>>>>>>> does the
>> >>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>> >>>>>>>>>
>> >>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
>> >>>>>>>>> are
>> >>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
>> >>>>>>>>> All these
>> >>>>>>>>> define several runtime parameters, which in turn define how
>> >>>>>>>>> well/badly
>> >>>>>>>>> will the pipeline perform and how many resources might be
>> >>>>>>>>> needed. From
>> >>>>>>>>> my point of view, pure streaming should be the most resource
>> >>>>>>>>> demanding
>> >>>>>>>>> (if not, why bother with batch? why not run everything in
>> >>>>>>>>> streaming
>> >>>>>>>>> only? what will there remain to "unify"?).
>> >>>>>>>>>
>> >>>>>>>>> > Fortunately, for batch, only the state for a single key
>> >>>>>>>>> needs to be
>> >>>>>>>>> preserved at a time, rather than the state for all keys across
>> >>>>>>>>> the range
>> >>>>>>>>> of skew. Of course if you have few or hot keys, one can still
>> >>>>>>>>> have
>> >>>>>>>>> issues (and this is not specific to StatefulDoFns).
>> >>>>>>>>>
>> >>>>>>>>> Yes, but here is still the presumption that my stateful DoFn
can
>> >>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
>> >>>>>>>>> case in
>> >>>>>>>>> more detail.
>> >>>>>>>>>
>> >>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
>> >>>>>>>>> some key for
>> >>>>>>>>> each element, which is irrelevant for the demonstration). Your
>> >>>>>>>>> task is
>> >>>>>>>>> to calculate in running global window the actual number of
>> >>>>>>>>> changes
>> >>>>>>>>> between state 0 and state 1 and vice versa. When the state
>> >>>>>>>>> doesn't
>> >>>>>>>>> change, you don't calculate anything. If input (for given key)
>> >>>>>>>>> would be
>> >>>>>>>>> (tN denotes timestamp N):
>> >>>>>>>>>
>> >>>>>>>>> t1: 1
>> >>>>>>>>>
>> >>>>>>>>> t2: 0
>> >>>>>>>>>
>> >>>>>>>>> t3: 0
>> >>>>>>>>>
>> >>>>>>>>> t4: 1
>> >>>>>>>>>
>> >>>>>>>>> t5: 1
>> >>>>>>>>>
>> >>>>>>>>> t6: 0
>> >>>>>>>>>
>> >>>>>>>>> then the output should yield (supposing that default state is
>> >>>>>>>>> zero):
>> >>>>>>>>>
>> >>>>>>>>> t1: (one: 1, zero: 0)
>> >>>>>>>>>
>> >>>>>>>>> t2: (one: 1, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t3: (one: 1, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t4: (one: 2, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t5: (one: 2, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t6: (one: 2, zero: 2)
>> >>>>>>>>>
>> >>>>>>>>> How would you implement this in current Beam semantics?
>> >>>>>>>> I think your saying here that I know that my input is ordered
>> >>>>>>>> in a specific way and since I assume the order when writing my
>> >>>>>>>> pipeline I can perform this optimization. But there is nothing
>> >>>>>>>> preventing a runner from noticing that your processing in the
>> >>>>>>>> global window with a specific type of trigger and re-ordering
>> >>>>>>>> your inputs/processing to get better performance (since you
>> >>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>> >>>>>>>> streaming for the GlobalWindow).
>> >>>>>>>>
>> >>>>>>>> Today, if you must have a strict order, you must guarantee that
>> >>>>>>>> your StatefulParDo implements the necessary "buffering &
>> >>>>>>>> sorting" into state. I can see why you would want an annotation
>> >>>>>>>> that says I must have timestamp ordered elements, since it
>> >>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
>> >>>>>>>> is a low-level function, it really is the "here you go and do
>> >>>>>>>> whatever you need to but here be dragons" function while
>> >>>>>>>> windowing and triggering is meant to keep many people from
>> >>>>>>>> writing StatefulParDo in the first place.
>> >>>>>>>>
>> >>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
>> >>>>>>>>> are likely to
>> >>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>> >>>>>>>>> falls
>> >>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>
>> >>>>>>>>> But the worst case is defined by input of size (available
>> >>>>>>>>> resources +
>> >>>>>>>>> single byte) -> pipeline fail. Although it could have
>> >>>>>>>>> finished, given
>> >>>>>>>>> the right conditions.
>> >>>>>>>>>
>> >>>>>>>>> > This might be reasonable, implemented by default by
>> >>>>>>>>> buffering
>> >>>>>>>>> everything and releasing elements as the watermark (+lateness)
>> >>>>>>>>> advances,
>> >>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
>> >>>>>>>>> reason
>> >>>>>>>>> about) code.
>> >>>>>>>>>
>> >>>>>>>>> Sure, the pipeline will be less efficient, because it would
>> >>>>>>>>> have to
>> >>>>>>>>> buffer and sort the inputs. But at least it will produce
>> >>>>>>>>> correct results
>> >>>>>>>>> in cases where updates to state are order-sensitive.
>> >>>>>>>>>
>> >>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
>> >>>>>>>>> (key, values):
>> >>>>>>>>> [(key, value) for value in values])?
>> >>>>>>>>>
>> >>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
>> >>>>>>>>> The trigger
>> >>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
>> >>>>>>>>> element
>> >>>>>>>>> with lowest timestamp in the buffer. Although this could be
>> >>>>>>>>> somehow
>> >>>>>>>>> emulated by fixed trigger each X millis.
>> >>>>>>>>>
>> >>>>>>>>> > Or is the underlying desire just to be able to hint to
>> >>>>>>>>> the runner
>> >>>>>>>>> that the code may perform better (e.g. require less resources)
>> >>>>>>>>> as skew
>> >>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
>> >>>>>>>>>
>> >>>>>>>>> No, the sorting would have to be done in streaming case as
>> >>>>>>>>> well. That is
>> >>>>>>>>> an imperative of the unified model. I think it is possible to
>> >>>>>>>>> sort by
>> >>>>>>>>> timestamp only in batch case (and do it for *all* batch
>> >>>>>>>>> stateful pardos
>> >>>>>>>>> without annotation), or introduce annotation, but then make
>> >>>>>>>>> the same
>> >>>>>>>>> guarantees for streaming case as well.
>> >>>>>>>>>
>> >>>>>>>>> Jan
>> >>>>>>>>>
>> >>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>> >>>>>>>>>> <je.ik@seznam.cz(mailto:je.ik@seznam.cz)> wrote:
>> >>>>>>>>>>> Hi Robert,
>> >>>>>>>>>>>
>> >>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
>> >>>>>>>>>>> guarantees
>> >>>>>>>>>>> of ordering are given in either mode, there is *implicit*
>> >>>>>>>>>>> ordering in
>> >>>>>>>>>>> streaming case that is due to nature of the processing - the
>> >>>>>>>>>>> difference
>> >>>>>>>>>>> between watermark and timestamp of elements flowing through
>> >>>>>>>>>>> the pipeline
>> >>>>>>>>>>> are generally low (too high difference leads to the
>> >>>>>>>>>>> overbuffering
>> >>>>>>>>>>> problem), but there is no such bound on batch.
>> >>>>>>>>>> Fortunately, for batch, only the state for a single key needs
>> >>>>>>>>>> to be
>> >>>>>>>>>> preserved at a time, rather than the state for all keys
>> >>>>>>>>>> across the
>> >>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
>> >>>>>>>>>> still
>> >>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>> >>>>>>>>>>
>> >>>>>>>>>>> As a result, I see a few possible solutions:
>> >>>>>>>>>>>
>> >>>>>>>>>>> - the best and most natural seems to be extension of
>> >>>>>>>>>>> the model, so
>> >>>>>>>>>>> that it defines batch as not only "streaming pipeline
>> >>>>>>>>>>> executed in batch
>> >>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>> >>>>>>>>>>> characteristics as
>> >>>>>>>>>>> in streaming case, executed in batch fashion", I really
>> >>>>>>>>>>> don't think that
>> >>>>>>>>>>> there are any conflicts with the current model, or that this
>> >>>>>>>>>>> could
>> >>>>>>>>>>> affect performance, because the required sorting (as pointed
by
>> >>>>>>>>>>> Aljoscha) is very probably already done during translation
>> >>>>>>>>>>> of stateful
>> >>>>>>>>>>> pardos. Also note that this definition only affects user
>> >>>>>>>>>>> defined
>> >>>>>>>>>>> stateful pardos
>> >>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
>> >>>>>>>>>> have
>> >>>>>>>>>> microbatch, or even a runner that alternates between
>> >>>>>>>>>> different modes.
>> >>>>>>>>>> The model describes what the valid outputs are given a
>> >>>>>>>>>> (sometimes
>> >>>>>>>>>> partial) set of inputs. It becomes really hard to define
>> >>>>>>>>>> things like
>> >>>>>>>>>> "as good runtime characteristics." Once you allow any
>> >>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
>> >>>>>>>>>> (and
>> >>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>> >>>>>>>>>> out-of-orderedness.
>> >>>>>>>>>>
>> >>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
>> >>>>>>>>>> likely to
>> >>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>> >>>>>>>>>> falls
>> >>>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>>
>> >>>>>>>>>>> - another option would be to introduce annotation for
>> >>>>>>>>>>> DoFns (e.g.
>> >>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
>> >>>>>>>>>>> the sorting
>> >>>>>>>>>>> in batch case - but - this extension would have to ensure
>> >>>>>>>>>>> the sorting in
>> >>>>>>>>>>> streaming mode also - it would require definition of allowed
>> >>>>>>>>>>> lateness,
>> >>>>>>>>>>> and triggger (essentially similar to window)
>> >>>>>>>>>> This might be reasonable, implemented by default by buffering
>> >>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>> >>>>>>>>>> advances, but would likely lead to inefficient (though
>> >>>>>>>>>> *maybe* easier
>> >>>>>>>>>> to reason about) code. Not sure about the semantics of
>> >>>>>>>>>> triggering
>> >>>>>>>>>> here, especially data-driven triggers. Would it be roughly
>> >>>>>>>>>> equivalent
>> >>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
>> >>>>>>>>>> value in
>> >>>>>>>>>> values])?
>> >>>>>>>>>>
>> >>>>>>>>>> Or is the underlying desire just to be able to hint to the
>> >>>>>>>>>> runner that
>> >>>>>>>>>> the code may perform better (e.g. require less resources) as
>> >>>>>>>>>> skew is
>> >>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>> >>>>>>>>>>
>> >>>>>>>>>>> - last option would be to introduce these "higher order
>> >>>>>>>>>>> guarantees" in
>> >>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
>> >>>>>>>>>>> worst
>> >>>>>>>>>>> option to me
>> >>>>>>>>>>>
>> >>>>>>>>>>> I see the first two options quite equally good, although the
>> >>>>>>>>>>> letter one
>> >>>>>>>>>>> is probably more time consuming to implement. But it would
>> >>>>>>>>>>> bring
>> >>>>>>>>>>> additional feature to streaming case as well.
>> >>>>>>>>>>>
>> >>>>>>>>>>> Thanks for any thoughts.
>> >>>>>>>>>>>
>> >>>>>>>>>>> Jan
>> >>>>>>>>>>>
>> >>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>> >>>>>>>>>>>> <je.ik@seznam.cz(mailto:je.ik@seznam.cz)> wrote:
>> >>>>>>>>>>>>> Hi Reuven,
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
>> >>>>>>>>>>>>>> runners.
>> >>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
>> >>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
>> >>>>>>>>>>>>> of elements. That basically (practically) can work only
>> >>>>>>>>>>>>> for cases, where the order of input elements doesn't
>> >>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
>> >>>>>>>>>>>>> time you have a state machine involved, then the ordering
>> >>>>>>>>>>>>> of elements would matter.
>> >>>>>>>>>>>> No guarantees on order are provided in *either* streaming
>> >>>>>>>>>>>> or batch
>> >>>>>>>>>>>> mode by the model. However, it is the case that in order to
>> >>>>>>>>>>>> make
>> >>>>>>>>>>>> forward progress most streaming runners attempt to limit
>> >>>>>>>>>>>> the amount of
>> >>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
>> >>>>>>>>>>>> processing
>> >>>>>>>>>>>> time) to make forward progress, which in turn could help
>> >>>>>>>>>>>> cap the
>> >>>>>>>>>>>> amount of state that must be held concurrently, whereas a
>> >>>>>>>>>>>> batch runner
>> >>>>>>>>>>>> may not allow any state to be safely discarded until the
whole
>> >>>>>>>>>>>> timeline from infinite past to infinite future has been
>> >>>>>>>>>>>> observed.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
>> >>>>>>>>>>>> batch" in batch mode.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>> >>>>>>>>>>>> <mxm@apache.org(mailto:mxm@apache.org)> wrote:
>> >>>>>>>>>>>>
>> >>>>>>>>>>>>>> batch semantics and streaming semantics differs only
>> >>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
>> >>>>>>>>>>>>>> batch and cannot on stream
>> >>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
>> >>>>>>>>>>>>> trigger. You
>> >>>>>>>>>>>>> could define additional triggers that do early firings.
>> >>>>>>>>>>>>> And you could
>> >>>>>>>>>>>>> even trigger the global window by advancing the watermark
>> >>>>>>>>>>>>> to +inf.
>> >>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
>> >>>>>>>>>>>> default
>> >>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
>> >>>>>>>>>>>> is more
>> >>>>>>>>>>>> likely to be user error than an actual desire to have no
>> >>>>>>>>>>>> output until
>> >>>>>>>>>>>> drain. But it's semantically valid in the model.
"
"
Re: Definition of Unified model
Posted by Reuven Lax <re...@google.com>.
So an example would be elements of type "startUserSession" and
"endUserSession" (website sessions, not Beam sessions). Logically you may
need to process them in the correct order if you have any sort of
state-machine logic. However timestamp ordering is never guaranteed to
match the logical ordering. Not only might you have several elements with
the same timestamp, but in reality time skew across backend servers can
cause the events to have timestamps in reverse order of the actual
causality order.
People do solve this problem today though. Publish the events to Kafka,
making sure that events for the same user end up in the same Kafka
partition. This ensures that the events appear in the Kafka partitions in
causality order, even if the timestamp order doesn't match. The your Kafka
subscriber simply process the elements in each partition in order.
I think the ability to impose FIFO causality ordering is what's needed for
any state-machine work. Timestamp ordering has advantages (though often I
think the advantage is in state), but does not solve this problem.
Reuven
On Thu, May 23, 2019 at 7:48 AM Robert Bradshaw <ro...@google.com> wrote:
> Good point.
>
> The "implementation-specific" way I would do this is
> window-by-instant, followed by a DoFn that gets all the elements with
> the same timestamp and sorts/acts accordingly, but this counts on the
> runner producing windows in timestamp order (likely?) and also the
> subsequent DoFn getting them in this order (also likely, due to
> fusion).
>
> One could make the argument that, though it does not provide
> deterministic behavior, getting elements of the same timestamp in
> different orders should produce equally valid interpretations of the
> data. (After all, due to relatively, timestamps are not technically
> well ordered across space.) I can see how data-dependent tiebreakers
> could be useful, or promises of preservation of order between
> operations.
>
> - Robert
>
> On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com> wrote:
> >
> > So Jan's example of state machines is quite a valid use case for
> ordering. However in my experience, timestamp ordering is insufficient for
> state machines. Elements that cause state transitions might come in with
> the exact same timestamp, yet still have a necessary ordering. Especially
> given Beam's decision to have milliseconds timestamps this is possible, but
> even at microsecond or nanosecond precision this can happen at scale. To
> handle state machines you usually need some sort of FIFO ordering along
> with an ordered sources, such as Kafka, not timestamp ordering.
> >
> > Reuven
> >
> > On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz> wrote:
> >>
> >> Hi all,
> >>
> >> thanks everyone for this discussion. I think I have gathered enough
> >> feedback to be able to put down a proposition for changes, which I will
> >> do and send to this list for further discussion. There are still doubts
> >> remaining the non-determinism and it's relation to outputs stability vs.
> >> latency. But I will try to clarify all this in the design document.
> >>
> >> Thanks,
> >>
> >> Jan
> >>
> >> On 5/22/19 3:49 PM, Maximilian Michels wrote:
> >> >> Someone from Flink might correct me if I'm wrong, but that's my
> >> >> current understanding.
> >> >
> >> > In essence your description of how exactly-once works in Flink is
> >> > correct. The general assumption in Flink is that pipelines must be
> >> > deterministic and thus produce idempotent writes in the case of
> >> > failures. However, that doesn't mean Beam sinks can't guarantee a bit
> >> > more with what Flink has to offer.
> >> >
> >> > Luke already mentioned the design discussions for @RequiresStableInput
> >> > which ensures idempotent writes for non-deterministic pipelines. This
> >> > is not part of the model but an optional Beam feature.
> >> >
> >> > We recently implemented support for @RequiresStableInput in the Flink
> >> > Runner. Reuven mentioned the Flink checkpoint confirmation, which
> >> > allows us to buffer (and checkpoint) processed data and only emit it
> >> > once a Flink checkpoint has completed.
> >> >
> >> > Cheers,
> >> > Max
> >> >
> >> > On 21.05.19 16:49, Jan Lukavský wrote:
> >> >> Hi,
> >> >>
> >> >> > Actually, I think it is a larger (open) question whether exactly
> >> >> once is guaranteed by the model or whether runners are allowed to
> >> >> relax that. I would think, however, that sources correctly
> >> >> implemented should be idempotent when run atop an exactly once
> >> >> infrastructure such as Flink of Dataflow.
> >> >>
> >> >> I would assume, that the model basically inherits guarantees of
> >> >> underlying infrastructure. Because Flink does not work as you
> >> >> described (atomic commit of inputs, state and outputs), but rather a
> >> >> checkpoint mark is flowing through the DAG much like watermark and on
> >> >> failures operators are restored and data reprocessed, it (IMHO)
> >> >> implies, that you have exactly once everywhere in the DAG *but*
> >> >> sinks. That is because sinks cannot be restored to previous state,
> >> >> instead sinks are supposed to be idempotent in order for the exactly
> >> >> once to really work (or at least be able to commit outputs on
> >> >> checkpoint in sink). That implies that if you don't have sink that is
> >> >> able to commit outputs atomically on checkpoint, the pipeline
> >> >> execution should be deterministic upon retries, otherwise shadow
> >> >> writes from failed paths of the pipeline might appear.
> >> >>
> >> >> Someone from Flink might correct me if I'm wrong, but that's my
> >> >> current understanding.
> >> >>
> >> >> > Sounds like we should make this clearer.
> >> >>
> >> >> I meant that you are right that we must not in any thoughts we are
> >> >> having forget that streams are by definition out-of-order. That is
> >> >> property that we cannot change. But - that doesn't limit us from
> >> >> creating operator that presents the data to UDF as if the stream was
> >> >> ideally sorted. It can do that by introducing latency, of course.
> >> >>
> >> >> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
> >> >>> Reza: One could provide something like this as a utility class, but
> >> >>> one downside is that it is not scale invariant. It requires a tuning
> >> >>> parameter that, if to small, won't mitigate the problem, but if to
> >> >>> big, greatly increases latency. (Possibly one could define a dynamic
> >> >>> session-like window to solve this though...) It also might be harder
> >> >>> for runners that *can* cheaply present stuff in timestamp order to
> >> >>> optimize. (That and, in practice, our annotation-style process
> methods
> >> >>> don't lend themselves to easy composition.) I think it could work in
> >> >>> specific cases though.
> >> >>>
> >> >>> More inline below.
> >> >>>
> >> >>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz>
> wrote:
> >> >>>> Hi Robert,
> >> >>>>
> >> >>>> > Beam has an exactly-once model. If the data was consumed, state
> >> >>>> mutated, and outputs written downstream (these three are committed
> >> >>>> together atomically) it will not be replayed. That does not, of
> >> >>>> course,
> >> >>>> solve the non-determanism due to ordering (including the fact that
> two
> >> >>>> operations reading the same PCollection may view different
> ordering).
> >> >>>>
> >> >>>> I think what you describe is a property of a runner, not of the
> model,
> >> >>>> right? I think if I run my pipeline on Flink I will not get this
> >> >>>> atomicity, because although Flink uses also exactly-once model if
> >> >>>> might
> >> >>>> write outputs multiple times.
> >> >>> Actually, I think it is a larger (open) question whether exactly
> once
> >> >>> is guaranteed by the model or whether runners are allowed to relax
> >> >>> that. I would think, however, that sources correctly implemented
> >> >>> should be idempotent when run atop an exactly once infrastructure
> such
> >> >>> as Flink of Dataflow.
> >> >>>
> >> >>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
> >> >>>> received
> >> >>>> in a specific order? In the current model, it is not. Being able to
> >> >>>> read, handle, and produced out-of-order data, including late data,
> >> >>>> is a
> >> >>>> pretty fundamental property of distributed systems.
> >> >>>>
> >> >>>> Yes, absolutely. The argument here is not that Stateful ParDo
> should
> >> >>>> presume to receive elements in any order, but to _present_ it as
> >> >>>> such to
> >> >>>> the user @ProcessElement function.
> >> >>> Sounds like we should make this clearer.
> >> >>>
> >> >>>> > 2) Given that some operations are easier (or possibly only
> >> >>>> possible)
> >> >>>> to write when operating on ordered data, and that different runners
> >> >>>> may
> >> >>>> have (significantly) cheaper ways to provide this ordering than
> can be
> >> >>>> done by the user themselves, should we elevate this to a property
> of
> >> >>>> (Stateful?)DoFns that the runner can provide? I think a compelling
> >> >>>> argument can be made here that we should.
> >> >>>>
> >> >>>> +1
> >> >>>>
> >> >>>> Jan
> >> >>>>
> >> >>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
> >> >>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >> >>>>>> > I don't see batch vs. streaming as part of the model. One
> >> >>>>>> can have
> >> >>>>>> microbatch, or even a runner that alternates between different
> >> >>>>>> modes.
> >> >>>>>>
> >> >>>>>> Although I understand motivation of this statement, this project
> >> >>>>>> name is
> >> >>>>>> "Apache Beam: An advanced unified programming model". What does
> the
> >> >>>>>> model unify, if "streaming vs. batch" is not part of the model?
> >> >>>>> What I mean is that streaming vs. batch is no longer part of the
> >> >>>>> model
> >> >>>>> (or ideally API), but pushed down to be a concern of the runner
> >> >>>>> (executor) of the pipeline.
> >> >>>>>
> >> >>>>>
> >> >>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz>
> >> >>>>> wrote:
> >> >>>>>> Hi Kenn,
> >> >>>>>>
> >> >>>>>> OK, so if we introduce annotation, we can have stateful ParDo
> >> >>>>>> with sorting, that would perfectly resolve my issues. I still
> >> >>>>>> have some doubts, though. Let me explain. The current behavior of
> >> >>>>>> stateful ParDo has the following properties:
> >> >>>>>>
> >> >>>>>> a) might fail in batch, although runs fine in streaming (that
> >> >>>>>> is due to the buffering, and unbounded lateness in batch, which
> >> >>>>>> was discussed back and forth in this thread)
> >> >>>>>>
> >> >>>>>> b) might be non deterministic (this is because the elements
> >> >>>>>> arrive at somewhat random order, and even if you do the operation
> >> >>>>>> "assign unique ID to elements" this might produce different
> >> >>>>>> results when run multiple times)
> >> >>>>> PCollections are *explicitly* unordered. Any operations that
> >> >>>>> assume or
> >> >>>>> depend on a specific ordering for correctness (or determinism)
> must
> >> >>>>> provide that ordering themselves (i.e. tolerate "arbitrary
> shuffling
> >> >>>>> of inputs"). As you point out, that may be very expensive if you
> have
> >> >>>>> very hot keys with very large (unbounded) timestamp skew.
> >> >>>>>
> >> >>>>> StatefulDoFns are low-level operations that should be used with
> care;
> >> >>>>> the simpler windowing model gives determinism in the face of
> >> >>>>> unordered
> >> >>>>> data (though late data and non-end-of-window triggering introduces
> >> >>>>> some of the non-determanism back in).
> >> >>>>>
> >> >>>>>> What worries me most is the property b), because it seems to me
> >> >>>>>> to have serious consequences - not only that if you run twice
> >> >>>>>> batch pipeline you would get different results, but even on
> >> >>>>>> streaming, when pipeline fails and gets restarted from
> >> >>>>>> checkpoint, produced output might differ from the previous run
> >> >>>>>> and data from the first run might have already been persisted
> >> >>>>>> into sink. That would create somewhat messy outputs.
> >> >>>>> Beam has an exactly-once model. If the data was consumed, state
> >> >>>>> mutated, and outputs written downstream (these three are committed
> >> >>>>> together atomically) it will not be replayed. That does not, of
> >> >>>>> course, solve the non-determanism due to ordering (including the
> fact
> >> >>>>> that two operations reading the same PCollection may view
> different
> >> >>>>> ordering).
> >> >>>>>
> >> >>>>>> These two properties makes me think that the current
> >> >>>>>> implementation is more of a _special case_ than the general one.
> >> >>>>>> The general one would be that your state doesn't have the
> >> >>>>>> properties to be able to tolerate buffering problems and/or
> >> >>>>>> non-determinism. Which is the case where you need sorting in both
> >> >>>>>> streaming and batch to be part of the model.
> >> >>>>>>
> >> >>>>>> Let me point out one more analogy - that is merging vs.
> >> >>>>>> non-merging windows. The general case (merging windows) implies
> >> >>>>>> sorting by timestamp in both batch case (explicit) and streaming
> >> >>>>>> (buffering). The special case (non-merging windows) doesn't rely
> >> >>>>>> on any timestamp ordering, so the sorting and buffering can be
> >> >>>>>> dropped. The underlying root cause of this is the same for both
> >> >>>>>> stateful ParDo and windowing (essentially, assigning window
> >> >>>>>> labels is a stateful operation when windowing function is
> merging).
> >> >>>>>>
> >> >>>>>> The reason for the current behavior of stateful ParDo seems to be
> >> >>>>>> performance, but is it right to abandon correctness in favor of
> >> >>>>>> performance? Wouldn't it be more consistent to have the default
> >> >>>>>> behavior prefer correctness and when you have the specific
> >> >>>>>> conditions of state function having special properties, then you
> >> >>>>>> can annotate your DoFn (with something like
> >> >>>>>> @TimeOrderingAgnostic), which would yield a better performance in
> >> >>>>>> that case?
> >> >>>>> There are two separable questions here.
> >> >>>>>
> >> >>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are
> received
> >> >>>>> in a specific order? In the current model, it is not. Being able
> to
> >> >>>>> read, handle, and produced out-of-order data, including late
> data, is
> >> >>>>> a pretty fundamental property of distributed systems.
> >> >>>>>
> >> >>>>> 2) Given that some operations are easier (or possibly only
> possible)
> >> >>>>> to write when operating on ordered data, and that different
> runners
> >> >>>>> may have (significantly) cheaper ways to provide this ordering
> than
> >> >>>>> can be done by the user themselves, should we elevate this to a
> >> >>>>> property of (Stateful?)DoFns that the runner can provide? I think
> a
> >> >>>>> compelling argument can be made here that we should.
> >> >>>>>
> >> >>>>> - Robert
> >> >>>>>
> >> >>>>>
> >> >>>>>
> >> >>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
> >> >>>>>>
> >> >>>>>> Thanks for the nice small example of a calculation that depends
> >> >>>>>> on order. You are right that many state machines have this
> >> >>>>>> property. I agree w/ you and Luke that it is convenient for batch
> >> >>>>>> processing to sort by event timestamp before running a stateful
> >> >>>>>> ParDo. In streaming you could also implement "sort by event
> >> >>>>>> timestamp" by buffering until you know all earlier data will be
> >> >>>>>> dropped - a slack buffer up to allowed lateness.
> >> >>>>>>
> >> >>>>>> I do not think that it is OK to sort in batch and not in
> >> >>>>>> streaming. Many state machines diverge very rapidly when things
> >> >>>>>> are out of order. So each runner if they see the
> >> >>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
> >> >>>>>> sorted data (by some mix of buffering and dropping), or to reject
> >> >>>>>> the pipeline as unsupported.
> >> >>>>>>
> >> >>>>>> And also want to say that this is not the default case - many
> >> >>>>>> uses of state & timers in ParDo yield different results at the
> >> >>>>>> element level, but the results are equivalent at in the big
> >> >>>>>> picture. Such as the example of "assign a unique sequence number
> >> >>>>>> to each element" or "group into batches" it doesn't matter
> >> >>>>>> exactly what the result is, only that it meets the spec. And
> >> >>>>>> other cases like user funnels are monotonic enough that you also
> >> >>>>>> don't actually need sorting.
> >> >>>>>>
> >> >>>>>> Kenn
> >> >>>>>>
> >> >>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz>
> >> >>>>>> wrote:
> >> >>>>>>> Yes, the problem will arise probably mostly when you have not
> >> >>>>>>> well distributed keys (or too few keys). I'm really not sure if
> >> >>>>>>> a pure GBK with a trigger can solve this - it might help to have
> >> >>>>>>> data driven trigger. There would still be some doubts, though.
> >> >>>>>>> The main question is still here - people say, that sorting by
> >> >>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
> >> >>>>>>> I don't really see why - the sorting is very probably already
> >> >>>>>>> there. And if not (hash grouping instead of sorted grouping),
> >> >>>>>>> then the sorting would affect only user defined StatefulParDos.
> >> >>>>>>>
> >> >>>>>>> This would suggest that the best way out of this would be really
> >> >>>>>>> to add annotation, so that the author of the pipeline can
> decide.
> >> >>>>>>>
> >> >>>>>>> If that would be acceptable I think I can try to prepare some
> >> >>>>>>> basic functionality, but I'm not sure, if I would be able to
> >> >>>>>>> cover all runners / sdks.
> >> >>>>>>>
> >> >>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
> >> >>>>>>>
> >> >>>>>>> It is read all per key and window and not just read all (this
> >> >>>>>>> still won't scale with hot keys in the global window). The GBK
> >> >>>>>>> preceding the StatefulParDo will guarantee that you are
> >> >>>>>>> processing all the values for a specific key and window at any
> >> >>>>>>> given time. Is there a specific window/trigger that is missing
> >> >>>>>>> that you feel would remove the need for you to use
> StatefulParDo?
> >> >>>>>>>
> >> >>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
> >> >>>>>>> wrote:
> >> >>>>>>>> Hi Lukasz,
> >> >>>>>>>>
> >> >>>>>>>>> Today, if you must have a strict order, you must guarantee
> >> >>>>>>>>> that your StatefulParDo implements the necessary "buffering &
> >> >>>>>>>>> sorting" into state.
> >> >>>>>>>> Yes, no problem with that. But this whole discussion started,
> >> >>>>>>>> because *this doesn't work on batch*. You simply cannot first
> >> >>>>>>>> read everything from distributed storage and then buffer it all
> >> >>>>>>>> into memory, just to read it again, but sorted. That will not
> >> >>>>>>>> work. And even if it would, it would be a terrible waste of
> >> >>>>>>>> resources.
> >> >>>>>>>>
> >> >>>>>>>> Jan
> >> >>>>>>>>
> >> >>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
> >> >>>>>>>>
> >> >>>>>>>>
> >> >>>>>>>>
> >> >>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
> >> >>>>>>>> wrote:
> >> >>>>>>>>> This discussion brings many really interesting questions for
> >> >>>>>>>>> me. :-)
> >> >>>>>>>>>
> >> >>>>>>>>> > I don't see batch vs. streaming as part of the model. One
> >> >>>>>>>>> can have
> >> >>>>>>>>> microbatch, or even a runner that alternates between different
> >> >>>>>>>>> modes.
> >> >>>>>>>>>
> >> >>>>>>>>> Although I understand motivation of this statement, this
> >> >>>>>>>>> project name is
> >> >>>>>>>>> "Apache Beam: An advanced unified programming model". What
> >> >>>>>>>>> does the
> >> >>>>>>>>> model unify, if "streaming vs. batch" is not part of the
> model?
> >> >>>>>>>>>
> >> >>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
> >> >>>>>>>>> are
> >> >>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
> >> >>>>>>>>> All these
> >> >>>>>>>>> define several runtime parameters, which in turn define how
> >> >>>>>>>>> well/badly
> >> >>>>>>>>> will the pipeline perform and how many resources might be
> >> >>>>>>>>> needed. From
> >> >>>>>>>>> my point of view, pure streaming should be the most resource
> >> >>>>>>>>> demanding
> >> >>>>>>>>> (if not, why bother with batch? why not run everything in
> >> >>>>>>>>> streaming
> >> >>>>>>>>> only? what will there remain to "unify"?).
> >> >>>>>>>>>
> >> >>>>>>>>> > Fortunately, for batch, only the state for a single key
> >> >>>>>>>>> needs to be
> >> >>>>>>>>> preserved at a time, rather than the state for all keys across
> >> >>>>>>>>> the range
> >> >>>>>>>>> of skew. Of course if you have few or hot keys, one can still
> >> >>>>>>>>> have
> >> >>>>>>>>> issues (and this is not specific to StatefulDoFns).
> >> >>>>>>>>>
> >> >>>>>>>>> Yes, but here is still the presumption that my stateful DoFn
> can
> >> >>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
> >> >>>>>>>>> case in
> >> >>>>>>>>> more detail.
> >> >>>>>>>>>
> >> >>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
> >> >>>>>>>>> some key for
> >> >>>>>>>>> each element, which is irrelevant for the demonstration). Your
> >> >>>>>>>>> task is
> >> >>>>>>>>> to calculate in running global window the actual number of
> >> >>>>>>>>> changes
> >> >>>>>>>>> between state 0 and state 1 and vice versa. When the state
> >> >>>>>>>>> doesn't
> >> >>>>>>>>> change, you don't calculate anything. If input (for given key)
> >> >>>>>>>>> would be
> >> >>>>>>>>> (tN denotes timestamp N):
> >> >>>>>>>>>
> >> >>>>>>>>> t1: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t2: 0
> >> >>>>>>>>>
> >> >>>>>>>>> t3: 0
> >> >>>>>>>>>
> >> >>>>>>>>> t4: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t5: 1
> >> >>>>>>>>>
> >> >>>>>>>>> t6: 0
> >> >>>>>>>>>
> >> >>>>>>>>> then the output should yield (supposing that default state is
> >> >>>>>>>>> zero):
> >> >>>>>>>>>
> >> >>>>>>>>> t1: (one: 1, zero: 0)
> >> >>>>>>>>>
> >> >>>>>>>>> t2: (one: 1, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t3: (one: 1, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t4: (one: 2, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t5: (one: 2, zero: 1)
> >> >>>>>>>>>
> >> >>>>>>>>> t6: (one: 2, zero: 2)
> >> >>>>>>>>>
> >> >>>>>>>>> How would you implement this in current Beam semantics?
> >> >>>>>>>> I think your saying here that I know that my input is ordered
> >> >>>>>>>> in a specific way and since I assume the order when writing my
> >> >>>>>>>> pipeline I can perform this optimization. But there is nothing
> >> >>>>>>>> preventing a runner from noticing that your processing in the
> >> >>>>>>>> global window with a specific type of trigger and re-ordering
> >> >>>>>>>> your inputs/processing to get better performance (since you
> >> >>>>>>>> can't use an AfterWatermark trigger for your pipeline in
> >> >>>>>>>> streaming for the GlobalWindow).
> >> >>>>>>>>
> >> >>>>>>>> Today, if you must have a strict order, you must guarantee that
> >> >>>>>>>> your StatefulParDo implements the necessary "buffering &
> >> >>>>>>>> sorting" into state. I can see why you would want an annotation
> >> >>>>>>>> that says I must have timestamp ordered elements, since it
> >> >>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
> >> >>>>>>>> is a low-level function, it really is the "here you go and do
> >> >>>>>>>> whatever you need to but here be dragons" function while
> >> >>>>>>>> windowing and triggering is meant to keep many people from
> >> >>>>>>>> writing StatefulParDo in the first place.
> >> >>>>>>>>
> >> >>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
> >> >>>>>>>>> are likely to
> >> >>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
> >> >>>>>>>>> falls
> >> >>>>>>>>> behind in streaming mode as well.
> >> >>>>>>>>>
> >> >>>>>>>>> But the worst case is defined by input of size (available
> >> >>>>>>>>> resources +
> >> >>>>>>>>> single byte) -> pipeline fail. Although it could have
> >> >>>>>>>>> finished, given
> >> >>>>>>>>> the right conditions.
> >> >>>>>>>>>
> >> >>>>>>>>> > This might be reasonable, implemented by default by
> >> >>>>>>>>> buffering
> >> >>>>>>>>> everything and releasing elements as the watermark (+lateness)
> >> >>>>>>>>> advances,
> >> >>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
> >> >>>>>>>>> reason
> >> >>>>>>>>> about) code.
> >> >>>>>>>>>
> >> >>>>>>>>> Sure, the pipeline will be less efficient, because it would
> >> >>>>>>>>> have to
> >> >>>>>>>>> buffer and sort the inputs. But at least it will produce
> >> >>>>>>>>> correct results
> >> >>>>>>>>> in cases where updates to state are order-sensitive.
> >> >>>>>>>>>
> >> >>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
> >> >>>>>>>>> (key, values):
> >> >>>>>>>>> [(key, value) for value in values])?
> >> >>>>>>>>>
> >> >>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
> >> >>>>>>>>> The trigger
> >> >>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
> >> >>>>>>>>> element
> >> >>>>>>>>> with lowest timestamp in the buffer. Although this could be
> >> >>>>>>>>> somehow
> >> >>>>>>>>> emulated by fixed trigger each X millis.
> >> >>>>>>>>>
> >> >>>>>>>>> > Or is the underlying desire just to be able to hint to
> >> >>>>>>>>> the runner
> >> >>>>>>>>> that the code may perform better (e.g. require less resources)
> >> >>>>>>>>> as skew
> >> >>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
> >> >>>>>>>>>
> >> >>>>>>>>> No, the sorting would have to be done in streaming case as
> >> >>>>>>>>> well. That is
> >> >>>>>>>>> an imperative of the unified model. I think it is possible to
> >> >>>>>>>>> sort by
> >> >>>>>>>>> timestamp only in batch case (and do it for *all* batch
> >> >>>>>>>>> stateful pardos
> >> >>>>>>>>> without annotation), or introduce annotation, but then make
> >> >>>>>>>>> the same
> >> >>>>>>>>> guarantees for streaming case as well.
> >> >>>>>>>>>
> >> >>>>>>>>> Jan
> >> >>>>>>>>>
> >> >>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
> >> >>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
> >> >>>>>>>>>> <je...@seznam.cz> wrote:
> >> >>>>>>>>>>> Hi Robert,
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
> >> >>>>>>>>>>> guarantees
> >> >>>>>>>>>>> of ordering are given in either mode, there is *implicit*
> >> >>>>>>>>>>> ordering in
> >> >>>>>>>>>>> streaming case that is due to nature of the processing - the
> >> >>>>>>>>>>> difference
> >> >>>>>>>>>>> between watermark and timestamp of elements flowing through
> >> >>>>>>>>>>> the pipeline
> >> >>>>>>>>>>> are generally low (too high difference leads to the
> >> >>>>>>>>>>> overbuffering
> >> >>>>>>>>>>> problem), but there is no such bound on batch.
> >> >>>>>>>>>> Fortunately, for batch, only the state for a single key needs
> >> >>>>>>>>>> to be
> >> >>>>>>>>>> preserved at a time, rather than the state for all keys
> >> >>>>>>>>>> across the
> >> >>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
> >> >>>>>>>>>> still
> >> >>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
> >> >>>>>>>>>>
> >> >>>>>>>>>>> As a result, I see a few possible solutions:
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> - the best and most natural seems to be extension of
> >> >>>>>>>>>>> the model, so
> >> >>>>>>>>>>> that it defines batch as not only "streaming pipeline
> >> >>>>>>>>>>> executed in batch
> >> >>>>>>>>>>> fashion", but "pipeline with at least as good runtime
> >> >>>>>>>>>>> characteristics as
> >> >>>>>>>>>>> in streaming case, executed in batch fashion", I really
> >> >>>>>>>>>>> don't think that
> >> >>>>>>>>>>> there are any conflicts with the current model, or that this
> >> >>>>>>>>>>> could
> >> >>>>>>>>>>> affect performance, because the required sorting (as
> pointed by
> >> >>>>>>>>>>> Aljoscha) is very probably already done during translation
> >> >>>>>>>>>>> of stateful
> >> >>>>>>>>>>> pardos. Also note that this definition only affects user
> >> >>>>>>>>>>> defined
> >> >>>>>>>>>>> stateful pardos
> >> >>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
> >> >>>>>>>>>> have
> >> >>>>>>>>>> microbatch, or even a runner that alternates between
> >> >>>>>>>>>> different modes.
> >> >>>>>>>>>> The model describes what the valid outputs are given a
> >> >>>>>>>>>> (sometimes
> >> >>>>>>>>>> partial) set of inputs. It becomes really hard to define
> >> >>>>>>>>>> things like
> >> >>>>>>>>>> "as good runtime characteristics." Once you allow any
> >> >>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
> >> >>>>>>>>>> (and
> >> >>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
> >> >>>>>>>>>> out-of-orderedness.
> >> >>>>>>>>>>
> >> >>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
> >> >>>>>>>>>> likely to
> >> >>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
> >> >>>>>>>>>> falls
> >> >>>>>>>>>> behind in streaming mode as well.
> >> >>>>>>>>>>
> >> >>>>>>>>>>> - another option would be to introduce annotation for
> >> >>>>>>>>>>> DoFns (e.g.
> >> >>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
> >> >>>>>>>>>>> the sorting
> >> >>>>>>>>>>> in batch case - but - this extension would have to ensure
> >> >>>>>>>>>>> the sorting in
> >> >>>>>>>>>>> streaming mode also - it would require definition of allowed
> >> >>>>>>>>>>> lateness,
> >> >>>>>>>>>>> and triggger (essentially similar to window)
> >> >>>>>>>>>> This might be reasonable, implemented by default by buffering
> >> >>>>>>>>>> everything and releasing elements as the watermark
> (+lateness)
> >> >>>>>>>>>> advances, but would likely lead to inefficient (though
> >> >>>>>>>>>> *maybe* easier
> >> >>>>>>>>>> to reason about) code. Not sure about the semantics of
> >> >>>>>>>>>> triggering
> >> >>>>>>>>>> here, especially data-driven triggers. Would it be roughly
> >> >>>>>>>>>> equivalent
> >> >>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
> >> >>>>>>>>>> value in
> >> >>>>>>>>>> values])?
> >> >>>>>>>>>>
> >> >>>>>>>>>> Or is the underlying desire just to be able to hint to the
> >> >>>>>>>>>> runner that
> >> >>>>>>>>>> the code may perform better (e.g. require less resources) as
> >> >>>>>>>>>> skew is
> >> >>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
> >> >>>>>>>>>>
> >> >>>>>>>>>>> - last option would be to introduce these "higher order
> >> >>>>>>>>>>> guarantees" in
> >> >>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
> >> >>>>>>>>>>> worst
> >> >>>>>>>>>>> option to me
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> I see the first two options quite equally good, although the
> >> >>>>>>>>>>> letter one
> >> >>>>>>>>>>> is probably more time consuming to implement. But it would
> >> >>>>>>>>>>> bring
> >> >>>>>>>>>>> additional feature to streaming case as well.
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> Thanks for any thoughts.
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> Jan
> >> >>>>>>>>>>>
> >> >>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
> >> >>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
> >> >>>>>>>>>>>> <je...@seznam.cz> wrote:
> >> >>>>>>>>>>>>> Hi Reuven,
> >> >>>>>>>>>>>>>
> >> >>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
> >> >>>>>>>>>>>>>> runners.
> >> >>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
> >> >>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
> >> >>>>>>>>>>>>> of elements. That basically (practically) can work only
> >> >>>>>>>>>>>>> for cases, where the order of input elements doesn't
> >> >>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
> >> >>>>>>>>>>>>> time you have a state machine involved, then the ordering
> >> >>>>>>>>>>>>> of elements would matter.
> >> >>>>>>>>>>>> No guarantees on order are provided in *either* streaming
> >> >>>>>>>>>>>> or batch
> >> >>>>>>>>>>>> mode by the model. However, it is the case that in order to
> >> >>>>>>>>>>>> make
> >> >>>>>>>>>>>> forward progress most streaming runners attempt to limit
> >> >>>>>>>>>>>> the amount of
> >> >>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
> >> >>>>>>>>>>>> processing
> >> >>>>>>>>>>>> time) to make forward progress, which in turn could help
> >> >>>>>>>>>>>> cap the
> >> >>>>>>>>>>>> amount of state that must be held concurrently, whereas a
> >> >>>>>>>>>>>> batch runner
> >> >>>>>>>>>>>> may not allow any state to be safely discarded until the
> whole
> >> >>>>>>>>>>>> timeline from infinite past to infinite future has been
> >> >>>>>>>>>>>> observed.
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
> >> >>>>>>>>>>>> batch" in batch mode.
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
> >> >>>>>>>>>>>> <mx...@apache.org> wrote:
> >> >>>>>>>>>>>>
> >> >>>>>>>>>>>>>> batch semantics and streaming semantics differs only
> >> >>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
> >> >>>>>>>>>>>>>> batch and cannot on stream
> >> >>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
> >> >>>>>>>>>>>>> trigger. You
> >> >>>>>>>>>>>>> could define additional triggers that do early firings.
> >> >>>>>>>>>>>>> And you could
> >> >>>>>>>>>>>>> even trigger the global window by advancing the watermark
> >> >>>>>>>>>>>>> to +inf.
> >> >>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
> >> >>>>>>>>>>>> default
> >> >>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
> >> >>>>>>>>>>>> is more
> >> >>>>>>>>>>>> likely to be user error than an actual desire to have no
> >> >>>>>>>>>>>> output until
> >> >>>>>>>>>>>> drain. But it's semantically valid in the model.
>
Re: Definition of Unified model
Posted by Robert Bradshaw <ro...@google.com>.
Good point.
The "implementation-specific" way I would do this is
window-by-instant, followed by a DoFn that gets all the elements with
the same timestamp and sorts/acts accordingly, but this counts on the
runner producing windows in timestamp order (likely?) and also the
subsequent DoFn getting them in this order (also likely, due to
fusion).
One could make the argument that, though it does not provide
deterministic behavior, getting elements of the same timestamp in
different orders should produce equally valid interpretations of the
data. (After all, due to relatively, timestamps are not technically
well ordered across space.) I can see how data-dependent tiebreakers
could be useful, or promises of preservation of order between
operations.
- Robert
On Thu, May 23, 2019 at 4:18 PM Reuven Lax <re...@google.com> wrote:
>
> So Jan's example of state machines is quite a valid use case for ordering. However in my experience, timestamp ordering is insufficient for state machines. Elements that cause state transitions might come in with the exact same timestamp, yet still have a necessary ordering. Especially given Beam's decision to have milliseconds timestamps this is possible, but even at microsecond or nanosecond precision this can happen at scale. To handle state machines you usually need some sort of FIFO ordering along with an ordered sources, such as Kafka, not timestamp ordering.
>
> Reuven
>
> On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz> wrote:
>>
>> Hi all,
>>
>> thanks everyone for this discussion. I think I have gathered enough
>> feedback to be able to put down a proposition for changes, which I will
>> do and send to this list for further discussion. There are still doubts
>> remaining the non-determinism and it's relation to outputs stability vs.
>> latency. But I will try to clarify all this in the design document.
>>
>> Thanks,
>>
>> Jan
>>
>> On 5/22/19 3:49 PM, Maximilian Michels wrote:
>> >> Someone from Flink might correct me if I'm wrong, but that's my
>> >> current understanding.
>> >
>> > In essence your description of how exactly-once works in Flink is
>> > correct. The general assumption in Flink is that pipelines must be
>> > deterministic and thus produce idempotent writes in the case of
>> > failures. However, that doesn't mean Beam sinks can't guarantee a bit
>> > more with what Flink has to offer.
>> >
>> > Luke already mentioned the design discussions for @RequiresStableInput
>> > which ensures idempotent writes for non-deterministic pipelines. This
>> > is not part of the model but an optional Beam feature.
>> >
>> > We recently implemented support for @RequiresStableInput in the Flink
>> > Runner. Reuven mentioned the Flink checkpoint confirmation, which
>> > allows us to buffer (and checkpoint) processed data and only emit it
>> > once a Flink checkpoint has completed.
>> >
>> > Cheers,
>> > Max
>> >
>> > On 21.05.19 16:49, Jan Lukavský wrote:
>> >> Hi,
>> >>
>> >> > Actually, I think it is a larger (open) question whether exactly
>> >> once is guaranteed by the model or whether runners are allowed to
>> >> relax that. I would think, however, that sources correctly
>> >> implemented should be idempotent when run atop an exactly once
>> >> infrastructure such as Flink of Dataflow.
>> >>
>> >> I would assume, that the model basically inherits guarantees of
>> >> underlying infrastructure. Because Flink does not work as you
>> >> described (atomic commit of inputs, state and outputs), but rather a
>> >> checkpoint mark is flowing through the DAG much like watermark and on
>> >> failures operators are restored and data reprocessed, it (IMHO)
>> >> implies, that you have exactly once everywhere in the DAG *but*
>> >> sinks. That is because sinks cannot be restored to previous state,
>> >> instead sinks are supposed to be idempotent in order for the exactly
>> >> once to really work (or at least be able to commit outputs on
>> >> checkpoint in sink). That implies that if you don't have sink that is
>> >> able to commit outputs atomically on checkpoint, the pipeline
>> >> execution should be deterministic upon retries, otherwise shadow
>> >> writes from failed paths of the pipeline might appear.
>> >>
>> >> Someone from Flink might correct me if I'm wrong, but that's my
>> >> current understanding.
>> >>
>> >> > Sounds like we should make this clearer.
>> >>
>> >> I meant that you are right that we must not in any thoughts we are
>> >> having forget that streams are by definition out-of-order. That is
>> >> property that we cannot change. But - that doesn't limit us from
>> >> creating operator that presents the data to UDF as if the stream was
>> >> ideally sorted. It can do that by introducing latency, of course.
>> >>
>> >> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>> >>> Reza: One could provide something like this as a utility class, but
>> >>> one downside is that it is not scale invariant. It requires a tuning
>> >>> parameter that, if to small, won't mitigate the problem, but if to
>> >>> big, greatly increases latency. (Possibly one could define a dynamic
>> >>> session-like window to solve this though...) It also might be harder
>> >>> for runners that *can* cheaply present stuff in timestamp order to
>> >>> optimize. (That and, in practice, our annotation-style process methods
>> >>> don't lend themselves to easy composition.) I think it could work in
>> >>> specific cases though.
>> >>>
>> >>> More inline below.
>> >>>
>> >>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz> wrote:
>> >>>> Hi Robert,
>> >>>>
>> >>>> > Beam has an exactly-once model. If the data was consumed, state
>> >>>> mutated, and outputs written downstream (these three are committed
>> >>>> together atomically) it will not be replayed. That does not, of
>> >>>> course,
>> >>>> solve the non-determanism due to ordering (including the fact that two
>> >>>> operations reading the same PCollection may view different ordering).
>> >>>>
>> >>>> I think what you describe is a property of a runner, not of the model,
>> >>>> right? I think if I run my pipeline on Flink I will not get this
>> >>>> atomicity, because although Flink uses also exactly-once model if
>> >>>> might
>> >>>> write outputs multiple times.
>> >>> Actually, I think it is a larger (open) question whether exactly once
>> >>> is guaranteed by the model or whether runners are allowed to relax
>> >>> that. I would think, however, that sources correctly implemented
>> >>> should be idempotent when run atop an exactly once infrastructure such
>> >>> as Flink of Dataflow.
>> >>>
>> >>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
>> >>>> received
>> >>>> in a specific order? In the current model, it is not. Being able to
>> >>>> read, handle, and produced out-of-order data, including late data,
>> >>>> is a
>> >>>> pretty fundamental property of distributed systems.
>> >>>>
>> >>>> Yes, absolutely. The argument here is not that Stateful ParDo should
>> >>>> presume to receive elements in any order, but to _present_ it as
>> >>>> such to
>> >>>> the user @ProcessElement function.
>> >>> Sounds like we should make this clearer.
>> >>>
>> >>>> > 2) Given that some operations are easier (or possibly only
>> >>>> possible)
>> >>>> to write when operating on ordered data, and that different runners
>> >>>> may
>> >>>> have (significantly) cheaper ways to provide this ordering than can be
>> >>>> done by the user themselves, should we elevate this to a property of
>> >>>> (Stateful?)DoFns that the runner can provide? I think a compelling
>> >>>> argument can be made here that we should.
>> >>>>
>> >>>> +1
>> >>>>
>> >>>> Jan
>> >>>>
>> >>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>> >>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz> wrote:
>> >>>>>> > I don't see batch vs. streaming as part of the model. One
>> >>>>>> can have
>> >>>>>> microbatch, or even a runner that alternates between different
>> >>>>>> modes.
>> >>>>>>
>> >>>>>> Although I understand motivation of this statement, this project
>> >>>>>> name is
>> >>>>>> "Apache Beam: An advanced unified programming model". What does the
>> >>>>>> model unify, if "streaming vs. batch" is not part of the model?
>> >>>>> What I mean is that streaming vs. batch is no longer part of the
>> >>>>> model
>> >>>>> (or ideally API), but pushed down to be a concern of the runner
>> >>>>> (executor) of the pipeline.
>> >>>>>
>> >>>>>
>> >>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz>
>> >>>>> wrote:
>> >>>>>> Hi Kenn,
>> >>>>>>
>> >>>>>> OK, so if we introduce annotation, we can have stateful ParDo
>> >>>>>> with sorting, that would perfectly resolve my issues. I still
>> >>>>>> have some doubts, though. Let me explain. The current behavior of
>> >>>>>> stateful ParDo has the following properties:
>> >>>>>>
>> >>>>>> a) might fail in batch, although runs fine in streaming (that
>> >>>>>> is due to the buffering, and unbounded lateness in batch, which
>> >>>>>> was discussed back and forth in this thread)
>> >>>>>>
>> >>>>>> b) might be non deterministic (this is because the elements
>> >>>>>> arrive at somewhat random order, and even if you do the operation
>> >>>>>> "assign unique ID to elements" this might produce different
>> >>>>>> results when run multiple times)
>> >>>>> PCollections are *explicitly* unordered. Any operations that
>> >>>>> assume or
>> >>>>> depend on a specific ordering for correctness (or determinism) must
>> >>>>> provide that ordering themselves (i.e. tolerate "arbitrary shuffling
>> >>>>> of inputs"). As you point out, that may be very expensive if you have
>> >>>>> very hot keys with very large (unbounded) timestamp skew.
>> >>>>>
>> >>>>> StatefulDoFns are low-level operations that should be used with care;
>> >>>>> the simpler windowing model gives determinism in the face of
>> >>>>> unordered
>> >>>>> data (though late data and non-end-of-window triggering introduces
>> >>>>> some of the non-determanism back in).
>> >>>>>
>> >>>>>> What worries me most is the property b), because it seems to me
>> >>>>>> to have serious consequences - not only that if you run twice
>> >>>>>> batch pipeline you would get different results, but even on
>> >>>>>> streaming, when pipeline fails and gets restarted from
>> >>>>>> checkpoint, produced output might differ from the previous run
>> >>>>>> and data from the first run might have already been persisted
>> >>>>>> into sink. That would create somewhat messy outputs.
>> >>>>> Beam has an exactly-once model. If the data was consumed, state
>> >>>>> mutated, and outputs written downstream (these three are committed
>> >>>>> together atomically) it will not be replayed. That does not, of
>> >>>>> course, solve the non-determanism due to ordering (including the fact
>> >>>>> that two operations reading the same PCollection may view different
>> >>>>> ordering).
>> >>>>>
>> >>>>>> These two properties makes me think that the current
>> >>>>>> implementation is more of a _special case_ than the general one.
>> >>>>>> The general one would be that your state doesn't have the
>> >>>>>> properties to be able to tolerate buffering problems and/or
>> >>>>>> non-determinism. Which is the case where you need sorting in both
>> >>>>>> streaming and batch to be part of the model.
>> >>>>>>
>> >>>>>> Let me point out one more analogy - that is merging vs.
>> >>>>>> non-merging windows. The general case (merging windows) implies
>> >>>>>> sorting by timestamp in both batch case (explicit) and streaming
>> >>>>>> (buffering). The special case (non-merging windows) doesn't rely
>> >>>>>> on any timestamp ordering, so the sorting and buffering can be
>> >>>>>> dropped. The underlying root cause of this is the same for both
>> >>>>>> stateful ParDo and windowing (essentially, assigning window
>> >>>>>> labels is a stateful operation when windowing function is merging).
>> >>>>>>
>> >>>>>> The reason for the current behavior of stateful ParDo seems to be
>> >>>>>> performance, but is it right to abandon correctness in favor of
>> >>>>>> performance? Wouldn't it be more consistent to have the default
>> >>>>>> behavior prefer correctness and when you have the specific
>> >>>>>> conditions of state function having special properties, then you
>> >>>>>> can annotate your DoFn (with something like
>> >>>>>> @TimeOrderingAgnostic), which would yield a better performance in
>> >>>>>> that case?
>> >>>>> There are two separable questions here.
>> >>>>>
>> >>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are received
>> >>>>> in a specific order? In the current model, it is not. Being able to
>> >>>>> read, handle, and produced out-of-order data, including late data, is
>> >>>>> a pretty fundamental property of distributed systems.
>> >>>>>
>> >>>>> 2) Given that some operations are easier (or possibly only possible)
>> >>>>> to write when operating on ordered data, and that different runners
>> >>>>> may have (significantly) cheaper ways to provide this ordering than
>> >>>>> can be done by the user themselves, should we elevate this to a
>> >>>>> property of (Stateful?)DoFns that the runner can provide? I think a
>> >>>>> compelling argument can be made here that we should.
>> >>>>>
>> >>>>> - Robert
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>> >>>>>>
>> >>>>>> Thanks for the nice small example of a calculation that depends
>> >>>>>> on order. You are right that many state machines have this
>> >>>>>> property. I agree w/ you and Luke that it is convenient for batch
>> >>>>>> processing to sort by event timestamp before running a stateful
>> >>>>>> ParDo. In streaming you could also implement "sort by event
>> >>>>>> timestamp" by buffering until you know all earlier data will be
>> >>>>>> dropped - a slack buffer up to allowed lateness.
>> >>>>>>
>> >>>>>> I do not think that it is OK to sort in batch and not in
>> >>>>>> streaming. Many state machines diverge very rapidly when things
>> >>>>>> are out of order. So each runner if they see the
>> >>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
>> >>>>>> sorted data (by some mix of buffering and dropping), or to reject
>> >>>>>> the pipeline as unsupported.
>> >>>>>>
>> >>>>>> And also want to say that this is not the default case - many
>> >>>>>> uses of state & timers in ParDo yield different results at the
>> >>>>>> element level, but the results are equivalent at in the big
>> >>>>>> picture. Such as the example of "assign a unique sequence number
>> >>>>>> to each element" or "group into batches" it doesn't matter
>> >>>>>> exactly what the result is, only that it meets the spec. And
>> >>>>>> other cases like user funnels are monotonic enough that you also
>> >>>>>> don't actually need sorting.
>> >>>>>>
>> >>>>>> Kenn
>> >>>>>>
>> >>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz>
>> >>>>>> wrote:
>> >>>>>>> Yes, the problem will arise probably mostly when you have not
>> >>>>>>> well distributed keys (or too few keys). I'm really not sure if
>> >>>>>>> a pure GBK with a trigger can solve this - it might help to have
>> >>>>>>> data driven trigger. There would still be some doubts, though.
>> >>>>>>> The main question is still here - people say, that sorting by
>> >>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
>> >>>>>>> I don't really see why - the sorting is very probably already
>> >>>>>>> there. And if not (hash grouping instead of sorted grouping),
>> >>>>>>> then the sorting would affect only user defined StatefulParDos.
>> >>>>>>>
>> >>>>>>> This would suggest that the best way out of this would be really
>> >>>>>>> to add annotation, so that the author of the pipeline can decide.
>> >>>>>>>
>> >>>>>>> If that would be acceptable I think I can try to prepare some
>> >>>>>>> basic functionality, but I'm not sure, if I would be able to
>> >>>>>>> cover all runners / sdks.
>> >>>>>>>
>> >>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>> >>>>>>>
>> >>>>>>> It is read all per key and window and not just read all (this
>> >>>>>>> still won't scale with hot keys in the global window). The GBK
>> >>>>>>> preceding the StatefulParDo will guarantee that you are
>> >>>>>>> processing all the values for a specific key and window at any
>> >>>>>>> given time. Is there a specific window/trigger that is missing
>> >>>>>>> that you feel would remove the need for you to use StatefulParDo?
>> >>>>>>>
>> >>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
>> >>>>>>> wrote:
>> >>>>>>>> Hi Lukasz,
>> >>>>>>>>
>> >>>>>>>>> Today, if you must have a strict order, you must guarantee
>> >>>>>>>>> that your StatefulParDo implements the necessary "buffering &
>> >>>>>>>>> sorting" into state.
>> >>>>>>>> Yes, no problem with that. But this whole discussion started,
>> >>>>>>>> because *this doesn't work on batch*. You simply cannot first
>> >>>>>>>> read everything from distributed storage and then buffer it all
>> >>>>>>>> into memory, just to read it again, but sorted. That will not
>> >>>>>>>> work. And even if it would, it would be a terrible waste of
>> >>>>>>>> resources.
>> >>>>>>>>
>> >>>>>>>> Jan
>> >>>>>>>>
>> >>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>> >>>>>>>>
>> >>>>>>>>
>> >>>>>>>>
>> >>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
>> >>>>>>>> wrote:
>> >>>>>>>>> This discussion brings many really interesting questions for
>> >>>>>>>>> me. :-)
>> >>>>>>>>>
>> >>>>>>>>> > I don't see batch vs. streaming as part of the model. One
>> >>>>>>>>> can have
>> >>>>>>>>> microbatch, or even a runner that alternates between different
>> >>>>>>>>> modes.
>> >>>>>>>>>
>> >>>>>>>>> Although I understand motivation of this statement, this
>> >>>>>>>>> project name is
>> >>>>>>>>> "Apache Beam: An advanced unified programming model". What
>> >>>>>>>>> does the
>> >>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>> >>>>>>>>>
>> >>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
>> >>>>>>>>> are
>> >>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
>> >>>>>>>>> All these
>> >>>>>>>>> define several runtime parameters, which in turn define how
>> >>>>>>>>> well/badly
>> >>>>>>>>> will the pipeline perform and how many resources might be
>> >>>>>>>>> needed. From
>> >>>>>>>>> my point of view, pure streaming should be the most resource
>> >>>>>>>>> demanding
>> >>>>>>>>> (if not, why bother with batch? why not run everything in
>> >>>>>>>>> streaming
>> >>>>>>>>> only? what will there remain to "unify"?).
>> >>>>>>>>>
>> >>>>>>>>> > Fortunately, for batch, only the state for a single key
>> >>>>>>>>> needs to be
>> >>>>>>>>> preserved at a time, rather than the state for all keys across
>> >>>>>>>>> the range
>> >>>>>>>>> of skew. Of course if you have few or hot keys, one can still
>> >>>>>>>>> have
>> >>>>>>>>> issues (and this is not specific to StatefulDoFns).
>> >>>>>>>>>
>> >>>>>>>>> Yes, but here is still the presumption that my stateful DoFn can
>> >>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
>> >>>>>>>>> case in
>> >>>>>>>>> more detail.
>> >>>>>>>>>
>> >>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
>> >>>>>>>>> some key for
>> >>>>>>>>> each element, which is irrelevant for the demonstration). Your
>> >>>>>>>>> task is
>> >>>>>>>>> to calculate in running global window the actual number of
>> >>>>>>>>> changes
>> >>>>>>>>> between state 0 and state 1 and vice versa. When the state
>> >>>>>>>>> doesn't
>> >>>>>>>>> change, you don't calculate anything. If input (for given key)
>> >>>>>>>>> would be
>> >>>>>>>>> (tN denotes timestamp N):
>> >>>>>>>>>
>> >>>>>>>>> t1: 1
>> >>>>>>>>>
>> >>>>>>>>> t2: 0
>> >>>>>>>>>
>> >>>>>>>>> t3: 0
>> >>>>>>>>>
>> >>>>>>>>> t4: 1
>> >>>>>>>>>
>> >>>>>>>>> t5: 1
>> >>>>>>>>>
>> >>>>>>>>> t6: 0
>> >>>>>>>>>
>> >>>>>>>>> then the output should yield (supposing that default state is
>> >>>>>>>>> zero):
>> >>>>>>>>>
>> >>>>>>>>> t1: (one: 1, zero: 0)
>> >>>>>>>>>
>> >>>>>>>>> t2: (one: 1, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t3: (one: 1, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t4: (one: 2, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t5: (one: 2, zero: 1)
>> >>>>>>>>>
>> >>>>>>>>> t6: (one: 2, zero: 2)
>> >>>>>>>>>
>> >>>>>>>>> How would you implement this in current Beam semantics?
>> >>>>>>>> I think your saying here that I know that my input is ordered
>> >>>>>>>> in a specific way and since I assume the order when writing my
>> >>>>>>>> pipeline I can perform this optimization. But there is nothing
>> >>>>>>>> preventing a runner from noticing that your processing in the
>> >>>>>>>> global window with a specific type of trigger and re-ordering
>> >>>>>>>> your inputs/processing to get better performance (since you
>> >>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>> >>>>>>>> streaming for the GlobalWindow).
>> >>>>>>>>
>> >>>>>>>> Today, if you must have a strict order, you must guarantee that
>> >>>>>>>> your StatefulParDo implements the necessary "buffering &
>> >>>>>>>> sorting" into state. I can see why you would want an annotation
>> >>>>>>>> that says I must have timestamp ordered elements, since it
>> >>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
>> >>>>>>>> is a low-level function, it really is the "here you go and do
>> >>>>>>>> whatever you need to but here be dragons" function while
>> >>>>>>>> windowing and triggering is meant to keep many people from
>> >>>>>>>> writing StatefulParDo in the first place.
>> >>>>>>>>
>> >>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
>> >>>>>>>>> are likely to
>> >>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>> >>>>>>>>> falls
>> >>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>
>> >>>>>>>>> But the worst case is defined by input of size (available
>> >>>>>>>>> resources +
>> >>>>>>>>> single byte) -> pipeline fail. Although it could have
>> >>>>>>>>> finished, given
>> >>>>>>>>> the right conditions.
>> >>>>>>>>>
>> >>>>>>>>> > This might be reasonable, implemented by default by
>> >>>>>>>>> buffering
>> >>>>>>>>> everything and releasing elements as the watermark (+lateness)
>> >>>>>>>>> advances,
>> >>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
>> >>>>>>>>> reason
>> >>>>>>>>> about) code.
>> >>>>>>>>>
>> >>>>>>>>> Sure, the pipeline will be less efficient, because it would
>> >>>>>>>>> have to
>> >>>>>>>>> buffer and sort the inputs. But at least it will produce
>> >>>>>>>>> correct results
>> >>>>>>>>> in cases where updates to state are order-sensitive.
>> >>>>>>>>>
>> >>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
>> >>>>>>>>> (key, values):
>> >>>>>>>>> [(key, value) for value in values])?
>> >>>>>>>>>
>> >>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
>> >>>>>>>>> The trigger
>> >>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
>> >>>>>>>>> element
>> >>>>>>>>> with lowest timestamp in the buffer. Although this could be
>> >>>>>>>>> somehow
>> >>>>>>>>> emulated by fixed trigger each X millis.
>> >>>>>>>>>
>> >>>>>>>>> > Or is the underlying desire just to be able to hint to
>> >>>>>>>>> the runner
>> >>>>>>>>> that the code may perform better (e.g. require less resources)
>> >>>>>>>>> as skew
>> >>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
>> >>>>>>>>>
>> >>>>>>>>> No, the sorting would have to be done in streaming case as
>> >>>>>>>>> well. That is
>> >>>>>>>>> an imperative of the unified model. I think it is possible to
>> >>>>>>>>> sort by
>> >>>>>>>>> timestamp only in batch case (and do it for *all* batch
>> >>>>>>>>> stateful pardos
>> >>>>>>>>> without annotation), or introduce annotation, but then make
>> >>>>>>>>> the same
>> >>>>>>>>> guarantees for streaming case as well.
>> >>>>>>>>>
>> >>>>>>>>> Jan
>> >>>>>>>>>
>> >>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>> >>>>>>>>>> <je...@seznam.cz> wrote:
>> >>>>>>>>>>> Hi Robert,
>> >>>>>>>>>>>
>> >>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
>> >>>>>>>>>>> guarantees
>> >>>>>>>>>>> of ordering are given in either mode, there is *implicit*
>> >>>>>>>>>>> ordering in
>> >>>>>>>>>>> streaming case that is due to nature of the processing - the
>> >>>>>>>>>>> difference
>> >>>>>>>>>>> between watermark and timestamp of elements flowing through
>> >>>>>>>>>>> the pipeline
>> >>>>>>>>>>> are generally low (too high difference leads to the
>> >>>>>>>>>>> overbuffering
>> >>>>>>>>>>> problem), but there is no such bound on batch.
>> >>>>>>>>>> Fortunately, for batch, only the state for a single key needs
>> >>>>>>>>>> to be
>> >>>>>>>>>> preserved at a time, rather than the state for all keys
>> >>>>>>>>>> across the
>> >>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
>> >>>>>>>>>> still
>> >>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>> >>>>>>>>>>
>> >>>>>>>>>>> As a result, I see a few possible solutions:
>> >>>>>>>>>>>
>> >>>>>>>>>>> - the best and most natural seems to be extension of
>> >>>>>>>>>>> the model, so
>> >>>>>>>>>>> that it defines batch as not only "streaming pipeline
>> >>>>>>>>>>> executed in batch
>> >>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>> >>>>>>>>>>> characteristics as
>> >>>>>>>>>>> in streaming case, executed in batch fashion", I really
>> >>>>>>>>>>> don't think that
>> >>>>>>>>>>> there are any conflicts with the current model, or that this
>> >>>>>>>>>>> could
>> >>>>>>>>>>> affect performance, because the required sorting (as pointed by
>> >>>>>>>>>>> Aljoscha) is very probably already done during translation
>> >>>>>>>>>>> of stateful
>> >>>>>>>>>>> pardos. Also note that this definition only affects user
>> >>>>>>>>>>> defined
>> >>>>>>>>>>> stateful pardos
>> >>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
>> >>>>>>>>>> have
>> >>>>>>>>>> microbatch, or even a runner that alternates between
>> >>>>>>>>>> different modes.
>> >>>>>>>>>> The model describes what the valid outputs are given a
>> >>>>>>>>>> (sometimes
>> >>>>>>>>>> partial) set of inputs. It becomes really hard to define
>> >>>>>>>>>> things like
>> >>>>>>>>>> "as good runtime characteristics." Once you allow any
>> >>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
>> >>>>>>>>>> (and
>> >>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>> >>>>>>>>>> out-of-orderedness.
>> >>>>>>>>>>
>> >>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
>> >>>>>>>>>> likely to
>> >>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>> >>>>>>>>>> falls
>> >>>>>>>>>> behind in streaming mode as well.
>> >>>>>>>>>>
>> >>>>>>>>>>> - another option would be to introduce annotation for
>> >>>>>>>>>>> DoFns (e.g.
>> >>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
>> >>>>>>>>>>> the sorting
>> >>>>>>>>>>> in batch case - but - this extension would have to ensure
>> >>>>>>>>>>> the sorting in
>> >>>>>>>>>>> streaming mode also - it would require definition of allowed
>> >>>>>>>>>>> lateness,
>> >>>>>>>>>>> and triggger (essentially similar to window)
>> >>>>>>>>>> This might be reasonable, implemented by default by buffering
>> >>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>> >>>>>>>>>> advances, but would likely lead to inefficient (though
>> >>>>>>>>>> *maybe* easier
>> >>>>>>>>>> to reason about) code. Not sure about the semantics of
>> >>>>>>>>>> triggering
>> >>>>>>>>>> here, especially data-driven triggers. Would it be roughly
>> >>>>>>>>>> equivalent
>> >>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
>> >>>>>>>>>> value in
>> >>>>>>>>>> values])?
>> >>>>>>>>>>
>> >>>>>>>>>> Or is the underlying desire just to be able to hint to the
>> >>>>>>>>>> runner that
>> >>>>>>>>>> the code may perform better (e.g. require less resources) as
>> >>>>>>>>>> skew is
>> >>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>> >>>>>>>>>>
>> >>>>>>>>>>> - last option would be to introduce these "higher order
>> >>>>>>>>>>> guarantees" in
>> >>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
>> >>>>>>>>>>> worst
>> >>>>>>>>>>> option to me
>> >>>>>>>>>>>
>> >>>>>>>>>>> I see the first two options quite equally good, although the
>> >>>>>>>>>>> letter one
>> >>>>>>>>>>> is probably more time consuming to implement. But it would
>> >>>>>>>>>>> bring
>> >>>>>>>>>>> additional feature to streaming case as well.
>> >>>>>>>>>>>
>> >>>>>>>>>>> Thanks for any thoughts.
>> >>>>>>>>>>>
>> >>>>>>>>>>> Jan
>> >>>>>>>>>>>
>> >>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>> >>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>> >>>>>>>>>>>> <je...@seznam.cz> wrote:
>> >>>>>>>>>>>>> Hi Reuven,
>> >>>>>>>>>>>>>
>> >>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
>> >>>>>>>>>>>>>> runners.
>> >>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
>> >>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
>> >>>>>>>>>>>>> of elements. That basically (practically) can work only
>> >>>>>>>>>>>>> for cases, where the order of input elements doesn't
>> >>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
>> >>>>>>>>>>>>> time you have a state machine involved, then the ordering
>> >>>>>>>>>>>>> of elements would matter.
>> >>>>>>>>>>>> No guarantees on order are provided in *either* streaming
>> >>>>>>>>>>>> or batch
>> >>>>>>>>>>>> mode by the model. However, it is the case that in order to
>> >>>>>>>>>>>> make
>> >>>>>>>>>>>> forward progress most streaming runners attempt to limit
>> >>>>>>>>>>>> the amount of
>> >>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
>> >>>>>>>>>>>> processing
>> >>>>>>>>>>>> time) to make forward progress, which in turn could help
>> >>>>>>>>>>>> cap the
>> >>>>>>>>>>>> amount of state that must be held concurrently, whereas a
>> >>>>>>>>>>>> batch runner
>> >>>>>>>>>>>> may not allow any state to be safely discarded until the whole
>> >>>>>>>>>>>> timeline from infinite past to infinite future has been
>> >>>>>>>>>>>> observed.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
>> >>>>>>>>>>>> batch" in batch mode.
>> >>>>>>>>>>>>
>> >>>>>>>>>>>>
>> >>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>> >>>>>>>>>>>> <mx...@apache.org> wrote:
>> >>>>>>>>>>>>
>> >>>>>>>>>>>>>> batch semantics and streaming semantics differs only
>> >>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
>> >>>>>>>>>>>>>> batch and cannot on stream
>> >>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
>> >>>>>>>>>>>>> trigger. You
>> >>>>>>>>>>>>> could define additional triggers that do early firings.
>> >>>>>>>>>>>>> And you could
>> >>>>>>>>>>>>> even trigger the global window by advancing the watermark
>> >>>>>>>>>>>>> to +inf.
>> >>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
>> >>>>>>>>>>>> default
>> >>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
>> >>>>>>>>>>>> is more
>> >>>>>>>>>>>> likely to be user error than an actual desire to have no
>> >>>>>>>>>>>> output until
>> >>>>>>>>>>>> drain. But it's semantically valid in the model.
Re: Definition of Unified model
Posted by Reuven Lax <re...@google.com>.
So Jan's example of state machines is quite a valid use case for ordering.
However in my experience, timestamp ordering is insufficient for state
machines. Elements that cause state transitions might come in with the
exact same timestamp, yet still have a necessary ordering. Especially given
Beam's decision to have milliseconds timestamps this is possible, but even
at microsecond or nanosecond precision this can happen at scale. To handle
state machines you usually need some sort of FIFO ordering along with an
ordered sources, such as Kafka, not timestamp ordering.
Reuven
On Thu, May 23, 2019 at 12:32 AM Jan Lukavský <je...@seznam.cz> wrote:
> Hi all,
>
> thanks everyone for this discussion. I think I have gathered enough
> feedback to be able to put down a proposition for changes, which I will
> do and send to this list for further discussion. There are still doubts
> remaining the non-determinism and it's relation to outputs stability vs.
> latency. But I will try to clarify all this in the design document.
>
> Thanks,
>
> Jan
>
> On 5/22/19 3:49 PM, Maximilian Michels wrote:
> >> Someone from Flink might correct me if I'm wrong, but that's my
> >> current understanding.
> >
> > In essence your description of how exactly-once works in Flink is
> > correct. The general assumption in Flink is that pipelines must be
> > deterministic and thus produce idempotent writes in the case of
> > failures. However, that doesn't mean Beam sinks can't guarantee a bit
> > more with what Flink has to offer.
> >
> > Luke already mentioned the design discussions for @RequiresStableInput
> > which ensures idempotent writes for non-deterministic pipelines. This
> > is not part of the model but an optional Beam feature.
> >
> > We recently implemented support for @RequiresStableInput in the Flink
> > Runner. Reuven mentioned the Flink checkpoint confirmation, which
> > allows us to buffer (and checkpoint) processed data and only emit it
> > once a Flink checkpoint has completed.
> >
> > Cheers,
> > Max
> >
> > On 21.05.19 16:49, Jan Lukavský wrote:
> >> Hi,
> >>
> >> > Actually, I think it is a larger (open) question whether exactly
> >> once is guaranteed by the model or whether runners are allowed to
> >> relax that. I would think, however, that sources correctly
> >> implemented should be idempotent when run atop an exactly once
> >> infrastructure such as Flink of Dataflow.
> >>
> >> I would assume, that the model basically inherits guarantees of
> >> underlying infrastructure. Because Flink does not work as you
> >> described (atomic commit of inputs, state and outputs), but rather a
> >> checkpoint mark is flowing through the DAG much like watermark and on
> >> failures operators are restored and data reprocessed, it (IMHO)
> >> implies, that you have exactly once everywhere in the DAG *but*
> >> sinks. That is because sinks cannot be restored to previous state,
> >> instead sinks are supposed to be idempotent in order for the exactly
> >> once to really work (or at least be able to commit outputs on
> >> checkpoint in sink). That implies that if you don't have sink that is
> >> able to commit outputs atomically on checkpoint, the pipeline
> >> execution should be deterministic upon retries, otherwise shadow
> >> writes from failed paths of the pipeline might appear.
> >>
> >> Someone from Flink might correct me if I'm wrong, but that's my
> >> current understanding.
> >>
> >> > Sounds like we should make this clearer.
> >>
> >> I meant that you are right that we must not in any thoughts we are
> >> having forget that streams are by definition out-of-order. That is
> >> property that we cannot change. But - that doesn't limit us from
> >> creating operator that presents the data to UDF as if the stream was
> >> ideally sorted. It can do that by introducing latency, of course.
> >>
> >> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
> >>> Reza: One could provide something like this as a utility class, but
> >>> one downside is that it is not scale invariant. It requires a tuning
> >>> parameter that, if to small, won't mitigate the problem, but if to
> >>> big, greatly increases latency. (Possibly one could define a dynamic
> >>> session-like window to solve this though...) It also might be harder
> >>> for runners that *can* cheaply present stuff in timestamp order to
> >>> optimize. (That and, in practice, our annotation-style process methods
> >>> don't lend themselves to easy composition.) I think it could work in
> >>> specific cases though.
> >>>
> >>> More inline below.
> >>>
> >>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz> wrote:
> >>>> Hi Robert,
> >>>>
> >>>> > Beam has an exactly-once model. If the data was consumed, state
> >>>> mutated, and outputs written downstream (these three are committed
> >>>> together atomically) it will not be replayed. That does not, of
> >>>> course,
> >>>> solve the non-determanism due to ordering (including the fact that two
> >>>> operations reading the same PCollection may view different ordering).
> >>>>
> >>>> I think what you describe is a property of a runner, not of the model,
> >>>> right? I think if I run my pipeline on Flink I will not get this
> >>>> atomicity, because although Flink uses also exactly-once model if
> >>>> might
> >>>> write outputs multiple times.
> >>> Actually, I think it is a larger (open) question whether exactly once
> >>> is guaranteed by the model or whether runners are allowed to relax
> >>> that. I would think, however, that sources correctly implemented
> >>> should be idempotent when run atop an exactly once infrastructure such
> >>> as Flink of Dataflow.
> >>>
> >>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
> >>>> received
> >>>> in a specific order? In the current model, it is not. Being able to
> >>>> read, handle, and produced out-of-order data, including late data,
> >>>> is a
> >>>> pretty fundamental property of distributed systems.
> >>>>
> >>>> Yes, absolutely. The argument here is not that Stateful ParDo should
> >>>> presume to receive elements in any order, but to _present_ it as
> >>>> such to
> >>>> the user @ProcessElement function.
> >>> Sounds like we should make this clearer.
> >>>
> >>>> > 2) Given that some operations are easier (or possibly only
> >>>> possible)
> >>>> to write when operating on ordered data, and that different runners
> >>>> may
> >>>> have (significantly) cheaper ways to provide this ordering than can be
> >>>> done by the user themselves, should we elevate this to a property of
> >>>> (Stateful?)DoFns that the runner can provide? I think a compelling
> >>>> argument can be made here that we should.
> >>>>
> >>>> +1
> >>>>
> >>>> Jan
> >>>>
> >>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
> >>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>> > I don't see batch vs. streaming as part of the model. One
> >>>>>> can have
> >>>>>> microbatch, or even a runner that alternates between different
> >>>>>> modes.
> >>>>>>
> >>>>>> Although I understand motivation of this statement, this project
> >>>>>> name is
> >>>>>> "Apache Beam: An advanced unified programming model". What does the
> >>>>>> model unify, if "streaming vs. batch" is not part of the model?
> >>>>> What I mean is that streaming vs. batch is no longer part of the
> >>>>> model
> >>>>> (or ideally API), but pushed down to be a concern of the runner
> >>>>> (executor) of the pipeline.
> >>>>>
> >>>>>
> >>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz>
> >>>>> wrote:
> >>>>>> Hi Kenn,
> >>>>>>
> >>>>>> OK, so if we introduce annotation, we can have stateful ParDo
> >>>>>> with sorting, that would perfectly resolve my issues. I still
> >>>>>> have some doubts, though. Let me explain. The current behavior of
> >>>>>> stateful ParDo has the following properties:
> >>>>>>
> >>>>>> a) might fail in batch, although runs fine in streaming (that
> >>>>>> is due to the buffering, and unbounded lateness in batch, which
> >>>>>> was discussed back and forth in this thread)
> >>>>>>
> >>>>>> b) might be non deterministic (this is because the elements
> >>>>>> arrive at somewhat random order, and even if you do the operation
> >>>>>> "assign unique ID to elements" this might produce different
> >>>>>> results when run multiple times)
> >>>>> PCollections are *explicitly* unordered. Any operations that
> >>>>> assume or
> >>>>> depend on a specific ordering for correctness (or determinism) must
> >>>>> provide that ordering themselves (i.e. tolerate "arbitrary shuffling
> >>>>> of inputs"). As you point out, that may be very expensive if you have
> >>>>> very hot keys with very large (unbounded) timestamp skew.
> >>>>>
> >>>>> StatefulDoFns are low-level operations that should be used with care;
> >>>>> the simpler windowing model gives determinism in the face of
> >>>>> unordered
> >>>>> data (though late data and non-end-of-window triggering introduces
> >>>>> some of the non-determanism back in).
> >>>>>
> >>>>>> What worries me most is the property b), because it seems to me
> >>>>>> to have serious consequences - not only that if you run twice
> >>>>>> batch pipeline you would get different results, but even on
> >>>>>> streaming, when pipeline fails and gets restarted from
> >>>>>> checkpoint, produced output might differ from the previous run
> >>>>>> and data from the first run might have already been persisted
> >>>>>> into sink. That would create somewhat messy outputs.
> >>>>> Beam has an exactly-once model. If the data was consumed, state
> >>>>> mutated, and outputs written downstream (these three are committed
> >>>>> together atomically) it will not be replayed. That does not, of
> >>>>> course, solve the non-determanism due to ordering (including the fact
> >>>>> that two operations reading the same PCollection may view different
> >>>>> ordering).
> >>>>>
> >>>>>> These two properties makes me think that the current
> >>>>>> implementation is more of a _special case_ than the general one.
> >>>>>> The general one would be that your state doesn't have the
> >>>>>> properties to be able to tolerate buffering problems and/or
> >>>>>> non-determinism. Which is the case where you need sorting in both
> >>>>>> streaming and batch to be part of the model.
> >>>>>>
> >>>>>> Let me point out one more analogy - that is merging vs.
> >>>>>> non-merging windows. The general case (merging windows) implies
> >>>>>> sorting by timestamp in both batch case (explicit) and streaming
> >>>>>> (buffering). The special case (non-merging windows) doesn't rely
> >>>>>> on any timestamp ordering, so the sorting and buffering can be
> >>>>>> dropped. The underlying root cause of this is the same for both
> >>>>>> stateful ParDo and windowing (essentially, assigning window
> >>>>>> labels is a stateful operation when windowing function is merging).
> >>>>>>
> >>>>>> The reason for the current behavior of stateful ParDo seems to be
> >>>>>> performance, but is it right to abandon correctness in favor of
> >>>>>> performance? Wouldn't it be more consistent to have the default
> >>>>>> behavior prefer correctness and when you have the specific
> >>>>>> conditions of state function having special properties, then you
> >>>>>> can annotate your DoFn (with something like
> >>>>>> @TimeOrderingAgnostic), which would yield a better performance in
> >>>>>> that case?
> >>>>> There are two separable questions here.
> >>>>>
> >>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are received
> >>>>> in a specific order? In the current model, it is not. Being able to
> >>>>> read, handle, and produced out-of-order data, including late data, is
> >>>>> a pretty fundamental property of distributed systems.
> >>>>>
> >>>>> 2) Given that some operations are easier (or possibly only possible)
> >>>>> to write when operating on ordered data, and that different runners
> >>>>> may have (significantly) cheaper ways to provide this ordering than
> >>>>> can be done by the user themselves, should we elevate this to a
> >>>>> property of (Stateful?)DoFns that the runner can provide? I think a
> >>>>> compelling argument can be made here that we should.
> >>>>>
> >>>>> - Robert
> >>>>>
> >>>>>
> >>>>>
> >>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
> >>>>>>
> >>>>>> Thanks for the nice small example of a calculation that depends
> >>>>>> on order. You are right that many state machines have this
> >>>>>> property. I agree w/ you and Luke that it is convenient for batch
> >>>>>> processing to sort by event timestamp before running a stateful
> >>>>>> ParDo. In streaming you could also implement "sort by event
> >>>>>> timestamp" by buffering until you know all earlier data will be
> >>>>>> dropped - a slack buffer up to allowed lateness.
> >>>>>>
> >>>>>> I do not think that it is OK to sort in batch and not in
> >>>>>> streaming. Many state machines diverge very rapidly when things
> >>>>>> are out of order. So each runner if they see the
> >>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
> >>>>>> sorted data (by some mix of buffering and dropping), or to reject
> >>>>>> the pipeline as unsupported.
> >>>>>>
> >>>>>> And also want to say that this is not the default case - many
> >>>>>> uses of state & timers in ParDo yield different results at the
> >>>>>> element level, but the results are equivalent at in the big
> >>>>>> picture. Such as the example of "assign a unique sequence number
> >>>>>> to each element" or "group into batches" it doesn't matter
> >>>>>> exactly what the result is, only that it meets the spec. And
> >>>>>> other cases like user funnels are monotonic enough that you also
> >>>>>> don't actually need sorting.
> >>>>>>
> >>>>>> Kenn
> >>>>>>
> >>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz>
> >>>>>> wrote:
> >>>>>>> Yes, the problem will arise probably mostly when you have not
> >>>>>>> well distributed keys (or too few keys). I'm really not sure if
> >>>>>>> a pure GBK with a trigger can solve this - it might help to have
> >>>>>>> data driven trigger. There would still be some doubts, though.
> >>>>>>> The main question is still here - people say, that sorting by
> >>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
> >>>>>>> I don't really see why - the sorting is very probably already
> >>>>>>> there. And if not (hash grouping instead of sorted grouping),
> >>>>>>> then the sorting would affect only user defined StatefulParDos.
> >>>>>>>
> >>>>>>> This would suggest that the best way out of this would be really
> >>>>>>> to add annotation, so that the author of the pipeline can decide.
> >>>>>>>
> >>>>>>> If that would be acceptable I think I can try to prepare some
> >>>>>>> basic functionality, but I'm not sure, if I would be able to
> >>>>>>> cover all runners / sdks.
> >>>>>>>
> >>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
> >>>>>>>
> >>>>>>> It is read all per key and window and not just read all (this
> >>>>>>> still won't scale with hot keys in the global window). The GBK
> >>>>>>> preceding the StatefulParDo will guarantee that you are
> >>>>>>> processing all the values for a specific key and window at any
> >>>>>>> given time. Is there a specific window/trigger that is missing
> >>>>>>> that you feel would remove the need for you to use StatefulParDo?
> >>>>>>>
> >>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
> >>>>>>> wrote:
> >>>>>>>> Hi Lukasz,
> >>>>>>>>
> >>>>>>>>> Today, if you must have a strict order, you must guarantee
> >>>>>>>>> that your StatefulParDo implements the necessary "buffering &
> >>>>>>>>> sorting" into state.
> >>>>>>>> Yes, no problem with that. But this whole discussion started,
> >>>>>>>> because *this doesn't work on batch*. You simply cannot first
> >>>>>>>> read everything from distributed storage and then buffer it all
> >>>>>>>> into memory, just to read it again, but sorted. That will not
> >>>>>>>> work. And even if it would, it would be a terrible waste of
> >>>>>>>> resources.
> >>>>>>>>
> >>>>>>>> Jan
> >>>>>>>>
> >>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
> >>>>>>>> wrote:
> >>>>>>>>> This discussion brings many really interesting questions for
> >>>>>>>>> me. :-)
> >>>>>>>>>
> >>>>>>>>> > I don't see batch vs. streaming as part of the model. One
> >>>>>>>>> can have
> >>>>>>>>> microbatch, or even a runner that alternates between different
> >>>>>>>>> modes.
> >>>>>>>>>
> >>>>>>>>> Although I understand motivation of this statement, this
> >>>>>>>>> project name is
> >>>>>>>>> "Apache Beam: An advanced unified programming model". What
> >>>>>>>>> does the
> >>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
> >>>>>>>>>
> >>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
> >>>>>>>>> are
> >>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
> >>>>>>>>> All these
> >>>>>>>>> define several runtime parameters, which in turn define how
> >>>>>>>>> well/badly
> >>>>>>>>> will the pipeline perform and how many resources might be
> >>>>>>>>> needed. From
> >>>>>>>>> my point of view, pure streaming should be the most resource
> >>>>>>>>> demanding
> >>>>>>>>> (if not, why bother with batch? why not run everything in
> >>>>>>>>> streaming
> >>>>>>>>> only? what will there remain to "unify"?).
> >>>>>>>>>
> >>>>>>>>> > Fortunately, for batch, only the state for a single key
> >>>>>>>>> needs to be
> >>>>>>>>> preserved at a time, rather than the state for all keys across
> >>>>>>>>> the range
> >>>>>>>>> of skew. Of course if you have few or hot keys, one can still
> >>>>>>>>> have
> >>>>>>>>> issues (and this is not specific to StatefulDoFns).
> >>>>>>>>>
> >>>>>>>>> Yes, but here is still the presumption that my stateful DoFn can
> >>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
> >>>>>>>>> case in
> >>>>>>>>> more detail.
> >>>>>>>>>
> >>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
> >>>>>>>>> some key for
> >>>>>>>>> each element, which is irrelevant for the demonstration). Your
> >>>>>>>>> task is
> >>>>>>>>> to calculate in running global window the actual number of
> >>>>>>>>> changes
> >>>>>>>>> between state 0 and state 1 and vice versa. When the state
> >>>>>>>>> doesn't
> >>>>>>>>> change, you don't calculate anything. If input (for given key)
> >>>>>>>>> would be
> >>>>>>>>> (tN denotes timestamp N):
> >>>>>>>>>
> >>>>>>>>> t1: 1
> >>>>>>>>>
> >>>>>>>>> t2: 0
> >>>>>>>>>
> >>>>>>>>> t3: 0
> >>>>>>>>>
> >>>>>>>>> t4: 1
> >>>>>>>>>
> >>>>>>>>> t5: 1
> >>>>>>>>>
> >>>>>>>>> t6: 0
> >>>>>>>>>
> >>>>>>>>> then the output should yield (supposing that default state is
> >>>>>>>>> zero):
> >>>>>>>>>
> >>>>>>>>> t1: (one: 1, zero: 0)
> >>>>>>>>>
> >>>>>>>>> t2: (one: 1, zero: 1)
> >>>>>>>>>
> >>>>>>>>> t3: (one: 1, zero: 1)
> >>>>>>>>>
> >>>>>>>>> t4: (one: 2, zero: 1)
> >>>>>>>>>
> >>>>>>>>> t5: (one: 2, zero: 1)
> >>>>>>>>>
> >>>>>>>>> t6: (one: 2, zero: 2)
> >>>>>>>>>
> >>>>>>>>> How would you implement this in current Beam semantics?
> >>>>>>>> I think your saying here that I know that my input is ordered
> >>>>>>>> in a specific way and since I assume the order when writing my
> >>>>>>>> pipeline I can perform this optimization. But there is nothing
> >>>>>>>> preventing a runner from noticing that your processing in the
> >>>>>>>> global window with a specific type of trigger and re-ordering
> >>>>>>>> your inputs/processing to get better performance (since you
> >>>>>>>> can't use an AfterWatermark trigger for your pipeline in
> >>>>>>>> streaming for the GlobalWindow).
> >>>>>>>>
> >>>>>>>> Today, if you must have a strict order, you must guarantee that
> >>>>>>>> your StatefulParDo implements the necessary "buffering &
> >>>>>>>> sorting" into state. I can see why you would want an annotation
> >>>>>>>> that says I must have timestamp ordered elements, since it
> >>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
> >>>>>>>> is a low-level function, it really is the "here you go and do
> >>>>>>>> whatever you need to but here be dragons" function while
> >>>>>>>> windowing and triggering is meant to keep many people from
> >>>>>>>> writing StatefulParDo in the first place.
> >>>>>>>>
> >>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
> >>>>>>>>> are likely to
> >>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
> >>>>>>>>> falls
> >>>>>>>>> behind in streaming mode as well.
> >>>>>>>>>
> >>>>>>>>> But the worst case is defined by input of size (available
> >>>>>>>>> resources +
> >>>>>>>>> single byte) -> pipeline fail. Although it could have
> >>>>>>>>> finished, given
> >>>>>>>>> the right conditions.
> >>>>>>>>>
> >>>>>>>>> > This might be reasonable, implemented by default by
> >>>>>>>>> buffering
> >>>>>>>>> everything and releasing elements as the watermark (+lateness)
> >>>>>>>>> advances,
> >>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
> >>>>>>>>> reason
> >>>>>>>>> about) code.
> >>>>>>>>>
> >>>>>>>>> Sure, the pipeline will be less efficient, because it would
> >>>>>>>>> have to
> >>>>>>>>> buffer and sort the inputs. But at least it will produce
> >>>>>>>>> correct results
> >>>>>>>>> in cases where updates to state are order-sensitive.
> >>>>>>>>>
> >>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
> >>>>>>>>> (key, values):
> >>>>>>>>> [(key, value) for value in values])?
> >>>>>>>>>
> >>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
> >>>>>>>>> The trigger
> >>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
> >>>>>>>>> element
> >>>>>>>>> with lowest timestamp in the buffer. Although this could be
> >>>>>>>>> somehow
> >>>>>>>>> emulated by fixed trigger each X millis.
> >>>>>>>>>
> >>>>>>>>> > Or is the underlying desire just to be able to hint to
> >>>>>>>>> the runner
> >>>>>>>>> that the code may perform better (e.g. require less resources)
> >>>>>>>>> as skew
> >>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
> >>>>>>>>>
> >>>>>>>>> No, the sorting would have to be done in streaming case as
> >>>>>>>>> well. That is
> >>>>>>>>> an imperative of the unified model. I think it is possible to
> >>>>>>>>> sort by
> >>>>>>>>> timestamp only in batch case (and do it for *all* batch
> >>>>>>>>> stateful pardos
> >>>>>>>>> without annotation), or introduce annotation, but then make
> >>>>>>>>> the same
> >>>>>>>>> guarantees for streaming case as well.
> >>>>>>>>>
> >>>>>>>>> Jan
> >>>>>>>>>
> >>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
> >>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
> >>>>>>>>>> <je...@seznam.cz> wrote:
> >>>>>>>>>>> Hi Robert,
> >>>>>>>>>>>
> >>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
> >>>>>>>>>>> guarantees
> >>>>>>>>>>> of ordering are given in either mode, there is *implicit*
> >>>>>>>>>>> ordering in
> >>>>>>>>>>> streaming case that is due to nature of the processing - the
> >>>>>>>>>>> difference
> >>>>>>>>>>> between watermark and timestamp of elements flowing through
> >>>>>>>>>>> the pipeline
> >>>>>>>>>>> are generally low (too high difference leads to the
> >>>>>>>>>>> overbuffering
> >>>>>>>>>>> problem), but there is no such bound on batch.
> >>>>>>>>>> Fortunately, for batch, only the state for a single key needs
> >>>>>>>>>> to be
> >>>>>>>>>> preserved at a time, rather than the state for all keys
> >>>>>>>>>> across the
> >>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
> >>>>>>>>>> still
> >>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
> >>>>>>>>>>
> >>>>>>>>>>> As a result, I see a few possible solutions:
> >>>>>>>>>>>
> >>>>>>>>>>> - the best and most natural seems to be extension of
> >>>>>>>>>>> the model, so
> >>>>>>>>>>> that it defines batch as not only "streaming pipeline
> >>>>>>>>>>> executed in batch
> >>>>>>>>>>> fashion", but "pipeline with at least as good runtime
> >>>>>>>>>>> characteristics as
> >>>>>>>>>>> in streaming case, executed in batch fashion", I really
> >>>>>>>>>>> don't think that
> >>>>>>>>>>> there are any conflicts with the current model, or that this
> >>>>>>>>>>> could
> >>>>>>>>>>> affect performance, because the required sorting (as pointed by
> >>>>>>>>>>> Aljoscha) is very probably already done during translation
> >>>>>>>>>>> of stateful
> >>>>>>>>>>> pardos. Also note that this definition only affects user
> >>>>>>>>>>> defined
> >>>>>>>>>>> stateful pardos
> >>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
> >>>>>>>>>> have
> >>>>>>>>>> microbatch, or even a runner that alternates between
> >>>>>>>>>> different modes.
> >>>>>>>>>> The model describes what the valid outputs are given a
> >>>>>>>>>> (sometimes
> >>>>>>>>>> partial) set of inputs. It becomes really hard to define
> >>>>>>>>>> things like
> >>>>>>>>>> "as good runtime characteristics." Once you allow any
> >>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
> >>>>>>>>>> (and
> >>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
> >>>>>>>>>> out-of-orderedness.
> >>>>>>>>>>
> >>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
> >>>>>>>>>> likely to
> >>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
> >>>>>>>>>> falls
> >>>>>>>>>> behind in streaming mode as well.
> >>>>>>>>>>
> >>>>>>>>>>> - another option would be to introduce annotation for
> >>>>>>>>>>> DoFns (e.g.
> >>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
> >>>>>>>>>>> the sorting
> >>>>>>>>>>> in batch case - but - this extension would have to ensure
> >>>>>>>>>>> the sorting in
> >>>>>>>>>>> streaming mode also - it would require definition of allowed
> >>>>>>>>>>> lateness,
> >>>>>>>>>>> and triggger (essentially similar to window)
> >>>>>>>>>> This might be reasonable, implemented by default by buffering
> >>>>>>>>>> everything and releasing elements as the watermark (+lateness)
> >>>>>>>>>> advances, but would likely lead to inefficient (though
> >>>>>>>>>> *maybe* easier
> >>>>>>>>>> to reason about) code. Not sure about the semantics of
> >>>>>>>>>> triggering
> >>>>>>>>>> here, especially data-driven triggers. Would it be roughly
> >>>>>>>>>> equivalent
> >>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
> >>>>>>>>>> value in
> >>>>>>>>>> values])?
> >>>>>>>>>>
> >>>>>>>>>> Or is the underlying desire just to be able to hint to the
> >>>>>>>>>> runner that
> >>>>>>>>>> the code may perform better (e.g. require less resources) as
> >>>>>>>>>> skew is
> >>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
> >>>>>>>>>>
> >>>>>>>>>>> - last option would be to introduce these "higher order
> >>>>>>>>>>> guarantees" in
> >>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
> >>>>>>>>>>> worst
> >>>>>>>>>>> option to me
> >>>>>>>>>>>
> >>>>>>>>>>> I see the first two options quite equally good, although the
> >>>>>>>>>>> letter one
> >>>>>>>>>>> is probably more time consuming to implement. But it would
> >>>>>>>>>>> bring
> >>>>>>>>>>> additional feature to streaming case as well.
> >>>>>>>>>>>
> >>>>>>>>>>> Thanks for any thoughts.
> >>>>>>>>>>>
> >>>>>>>>>>> Jan
> >>>>>>>>>>>
> >>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
> >>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
> >>>>>>>>>>>> <je...@seznam.cz> wrote:
> >>>>>>>>>>>>> Hi Reuven,
> >>>>>>>>>>>>>
> >>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
> >>>>>>>>>>>>>> runners.
> >>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
> >>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
> >>>>>>>>>>>>> of elements. That basically (practically) can work only
> >>>>>>>>>>>>> for cases, where the order of input elements doesn't
> >>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
> >>>>>>>>>>>>> time you have a state machine involved, then the ordering
> >>>>>>>>>>>>> of elements would matter.
> >>>>>>>>>>>> No guarantees on order are provided in *either* streaming
> >>>>>>>>>>>> or batch
> >>>>>>>>>>>> mode by the model. However, it is the case that in order to
> >>>>>>>>>>>> make
> >>>>>>>>>>>> forward progress most streaming runners attempt to limit
> >>>>>>>>>>>> the amount of
> >>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
> >>>>>>>>>>>> processing
> >>>>>>>>>>>> time) to make forward progress, which in turn could help
> >>>>>>>>>>>> cap the
> >>>>>>>>>>>> amount of state that must be held concurrently, whereas a
> >>>>>>>>>>>> batch runner
> >>>>>>>>>>>> may not allow any state to be safely discarded until the whole
> >>>>>>>>>>>> timeline from infinite past to infinite future has been
> >>>>>>>>>>>> observed.
> >>>>>>>>>>>>
> >>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
> >>>>>>>>>>>> batch" in batch mode.
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
> >>>>>>>>>>>> <mx...@apache.org> wrote:
> >>>>>>>>>>>>
> >>>>>>>>>>>>>> batch semantics and streaming semantics differs only
> >>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
> >>>>>>>>>>>>>> batch and cannot on stream
> >>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
> >>>>>>>>>>>>> trigger. You
> >>>>>>>>>>>>> could define additional triggers that do early firings.
> >>>>>>>>>>>>> And you could
> >>>>>>>>>>>>> even trigger the global window by advancing the watermark
> >>>>>>>>>>>>> to +inf.
> >>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
> >>>>>>>>>>>> default
> >>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
> >>>>>>>>>>>> is more
> >>>>>>>>>>>> likely to be user error than an actual desire to have no
> >>>>>>>>>>>> output until
> >>>>>>>>>>>> drain. But it's semantically valid in the model.
>
Re: Definition of Unified model
Posted by Jan Lukavský <je...@seznam.cz>.
Hi all,
thanks everyone for this discussion. I think I have gathered enough
feedback to be able to put down a proposition for changes, which I will
do and send to this list for further discussion. There are still doubts
remaining the non-determinism and it's relation to outputs stability vs.
latency. But I will try to clarify all this in the design document.
Thanks,
Jan
On 5/22/19 3:49 PM, Maximilian Michels wrote:
>> Someone from Flink might correct me if I'm wrong, but that's my
>> current understanding.
>
> In essence your description of how exactly-once works in Flink is
> correct. The general assumption in Flink is that pipelines must be
> deterministic and thus produce idempotent writes in the case of
> failures. However, that doesn't mean Beam sinks can't guarantee a bit
> more with what Flink has to offer.
>
> Luke already mentioned the design discussions for @RequiresStableInput
> which ensures idempotent writes for non-deterministic pipelines. This
> is not part of the model but an optional Beam feature.
>
> We recently implemented support for @RequiresStableInput in the Flink
> Runner. Reuven mentioned the Flink checkpoint confirmation, which
> allows us to buffer (and checkpoint) processed data and only emit it
> once a Flink checkpoint has completed.
>
> Cheers,
> Max
>
> On 21.05.19 16:49, Jan Lukavský wrote:
>> Hi,
>>
>> > Actually, I think it is a larger (open) question whether exactly
>> once is guaranteed by the model or whether runners are allowed to
>> relax that. I would think, however, that sources correctly
>> implemented should be idempotent when run atop an exactly once
>> infrastructure such as Flink of Dataflow.
>>
>> I would assume, that the model basically inherits guarantees of
>> underlying infrastructure. Because Flink does not work as you
>> described (atomic commit of inputs, state and outputs), but rather a
>> checkpoint mark is flowing through the DAG much like watermark and on
>> failures operators are restored and data reprocessed, it (IMHO)
>> implies, that you have exactly once everywhere in the DAG *but*
>> sinks. That is because sinks cannot be restored to previous state,
>> instead sinks are supposed to be idempotent in order for the exactly
>> once to really work (or at least be able to commit outputs on
>> checkpoint in sink). That implies that if you don't have sink that is
>> able to commit outputs atomically on checkpoint, the pipeline
>> execution should be deterministic upon retries, otherwise shadow
>> writes from failed paths of the pipeline might appear.
>>
>> Someone from Flink might correct me if I'm wrong, but that's my
>> current understanding.
>>
>> > Sounds like we should make this clearer.
>>
>> I meant that you are right that we must not in any thoughts we are
>> having forget that streams are by definition out-of-order. That is
>> property that we cannot change. But - that doesn't limit us from
>> creating operator that presents the data to UDF as if the stream was
>> ideally sorted. It can do that by introducing latency, of course.
>>
>> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>>> Reza: One could provide something like this as a utility class, but
>>> one downside is that it is not scale invariant. It requires a tuning
>>> parameter that, if to small, won't mitigate the problem, but if to
>>> big, greatly increases latency. (Possibly one could define a dynamic
>>> session-like window to solve this though...) It also might be harder
>>> for runners that *can* cheaply present stuff in timestamp order to
>>> optimize. (That and, in practice, our annotation-style process methods
>>> don't lend themselves to easy composition.) I think it could work in
>>> specific cases though.
>>>
>>> More inline below.
>>>
>>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>> Hi Robert,
>>>>
>>>> > Beam has an exactly-once model. If the data was consumed, state
>>>> mutated, and outputs written downstream (these three are committed
>>>> together atomically) it will not be replayed. That does not, of
>>>> course,
>>>> solve the non-determanism due to ordering (including the fact that two
>>>> operations reading the same PCollection may view different ordering).
>>>>
>>>> I think what you describe is a property of a runner, not of the model,
>>>> right? I think if I run my pipeline on Flink I will not get this
>>>> atomicity, because although Flink uses also exactly-once model if
>>>> might
>>>> write outputs multiple times.
>>> Actually, I think it is a larger (open) question whether exactly once
>>> is guaranteed by the model or whether runners are allowed to relax
>>> that. I would think, however, that sources correctly implemented
>>> should be idempotent when run atop an exactly once infrastructure such
>>> as Flink of Dataflow.
>>>
>>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
>>>> received
>>>> in a specific order? In the current model, it is not. Being able to
>>>> read, handle, and produced out-of-order data, including late data,
>>>> is a
>>>> pretty fundamental property of distributed systems.
>>>>
>>>> Yes, absolutely. The argument here is not that Stateful ParDo should
>>>> presume to receive elements in any order, but to _present_ it as
>>>> such to
>>>> the user @ProcessElement function.
>>> Sounds like we should make this clearer.
>>>
>>>> > 2) Given that some operations are easier (or possibly only
>>>> possible)
>>>> to write when operating on ordered data, and that different runners
>>>> may
>>>> have (significantly) cheaper ways to provide this ordering than can be
>>>> done by the user themselves, should we elevate this to a property of
>>>> (Stateful?)DoFns that the runner can provide? I think a compelling
>>>> argument can be made here that we should.
>>>>
>>>> +1
>>>>
>>>> Jan
>>>>
>>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz> wrote:
>>>>>> > I don't see batch vs. streaming as part of the model. One
>>>>>> can have
>>>>>> microbatch, or even a runner that alternates between different
>>>>>> modes.
>>>>>>
>>>>>> Although I understand motivation of this statement, this project
>>>>>> name is
>>>>>> "Apache Beam: An advanced unified programming model". What does the
>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>>>>> What I mean is that streaming vs. batch is no longer part of the
>>>>> model
>>>>> (or ideally API), but pushed down to be a concern of the runner
>>>>> (executor) of the pipeline.
>>>>>
>>>>>
>>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz>
>>>>> wrote:
>>>>>> Hi Kenn,
>>>>>>
>>>>>> OK, so if we introduce annotation, we can have stateful ParDo
>>>>>> with sorting, that would perfectly resolve my issues. I still
>>>>>> have some doubts, though. Let me explain. The current behavior of
>>>>>> stateful ParDo has the following properties:
>>>>>>
>>>>>> a) might fail in batch, although runs fine in streaming (that
>>>>>> is due to the buffering, and unbounded lateness in batch, which
>>>>>> was discussed back and forth in this thread)
>>>>>>
>>>>>> b) might be non deterministic (this is because the elements
>>>>>> arrive at somewhat random order, and even if you do the operation
>>>>>> "assign unique ID to elements" this might produce different
>>>>>> results when run multiple times)
>>>>> PCollections are *explicitly* unordered. Any operations that
>>>>> assume or
>>>>> depend on a specific ordering for correctness (or determinism) must
>>>>> provide that ordering themselves (i.e. tolerate "arbitrary shuffling
>>>>> of inputs"). As you point out, that may be very expensive if you have
>>>>> very hot keys with very large (unbounded) timestamp skew.
>>>>>
>>>>> StatefulDoFns are low-level operations that should be used with care;
>>>>> the simpler windowing model gives determinism in the face of
>>>>> unordered
>>>>> data (though late data and non-end-of-window triggering introduces
>>>>> some of the non-determanism back in).
>>>>>
>>>>>> What worries me most is the property b), because it seems to me
>>>>>> to have serious consequences - not only that if you run twice
>>>>>> batch pipeline you would get different results, but even on
>>>>>> streaming, when pipeline fails and gets restarted from
>>>>>> checkpoint, produced output might differ from the previous run
>>>>>> and data from the first run might have already been persisted
>>>>>> into sink. That would create somewhat messy outputs.
>>>>> Beam has an exactly-once model. If the data was consumed, state
>>>>> mutated, and outputs written downstream (these three are committed
>>>>> together atomically) it will not be replayed. That does not, of
>>>>> course, solve the non-determanism due to ordering (including the fact
>>>>> that two operations reading the same PCollection may view different
>>>>> ordering).
>>>>>
>>>>>> These two properties makes me think that the current
>>>>>> implementation is more of a _special case_ than the general one.
>>>>>> The general one would be that your state doesn't have the
>>>>>> properties to be able to tolerate buffering problems and/or
>>>>>> non-determinism. Which is the case where you need sorting in both
>>>>>> streaming and batch to be part of the model.
>>>>>>
>>>>>> Let me point out one more analogy - that is merging vs.
>>>>>> non-merging windows. The general case (merging windows) implies
>>>>>> sorting by timestamp in both batch case (explicit) and streaming
>>>>>> (buffering). The special case (non-merging windows) doesn't rely
>>>>>> on any timestamp ordering, so the sorting and buffering can be
>>>>>> dropped. The underlying root cause of this is the same for both
>>>>>> stateful ParDo and windowing (essentially, assigning window
>>>>>> labels is a stateful operation when windowing function is merging).
>>>>>>
>>>>>> The reason for the current behavior of stateful ParDo seems to be
>>>>>> performance, but is it right to abandon correctness in favor of
>>>>>> performance? Wouldn't it be more consistent to have the default
>>>>>> behavior prefer correctness and when you have the specific
>>>>>> conditions of state function having special properties, then you
>>>>>> can annotate your DoFn (with something like
>>>>>> @TimeOrderingAgnostic), which would yield a better performance in
>>>>>> that case?
>>>>> There are two separable questions here.
>>>>>
>>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are received
>>>>> in a specific order? In the current model, it is not. Being able to
>>>>> read, handle, and produced out-of-order data, including late data, is
>>>>> a pretty fundamental property of distributed systems.
>>>>>
>>>>> 2) Given that some operations are easier (or possibly only possible)
>>>>> to write when operating on ordered data, and that different runners
>>>>> may have (significantly) cheaper ways to provide this ordering than
>>>>> can be done by the user themselves, should we elevate this to a
>>>>> property of (Stateful?)DoFns that the runner can provide? I think a
>>>>> compelling argument can be made here that we should.
>>>>>
>>>>> - Robert
>>>>>
>>>>>
>>>>>
>>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>>>>>>
>>>>>> Thanks for the nice small example of a calculation that depends
>>>>>> on order. You are right that many state machines have this
>>>>>> property. I agree w/ you and Luke that it is convenient for batch
>>>>>> processing to sort by event timestamp before running a stateful
>>>>>> ParDo. In streaming you could also implement "sort by event
>>>>>> timestamp" by buffering until you know all earlier data will be
>>>>>> dropped - a slack buffer up to allowed lateness.
>>>>>>
>>>>>> I do not think that it is OK to sort in batch and not in
>>>>>> streaming. Many state machines diverge very rapidly when things
>>>>>> are out of order. So each runner if they see the
>>>>>> "@OrderByTimestamp" annotation (or whatever) needs to deliver
>>>>>> sorted data (by some mix of buffering and dropping), or to reject
>>>>>> the pipeline as unsupported.
>>>>>>
>>>>>> And also want to say that this is not the default case - many
>>>>>> uses of state & timers in ParDo yield different results at the
>>>>>> element level, but the results are equivalent at in the big
>>>>>> picture. Such as the example of "assign a unique sequence number
>>>>>> to each element" or "group into batches" it doesn't matter
>>>>>> exactly what the result is, only that it meets the spec. And
>>>>>> other cases like user funnels are monotonic enough that you also
>>>>>> don't actually need sorting.
>>>>>>
>>>>>> Kenn
>>>>>>
>>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz>
>>>>>> wrote:
>>>>>>> Yes, the problem will arise probably mostly when you have not
>>>>>>> well distributed keys (or too few keys). I'm really not sure if
>>>>>>> a pure GBK with a trigger can solve this - it might help to have
>>>>>>> data driven trigger. There would still be some doubts, though.
>>>>>>> The main question is still here - people say, that sorting by
>>>>>>> timestamp before stateful ParDo would be prohibitively slow, but
>>>>>>> I don't really see why - the sorting is very probably already
>>>>>>> there. And if not (hash grouping instead of sorted grouping),
>>>>>>> then the sorting would affect only user defined StatefulParDos.
>>>>>>>
>>>>>>> This would suggest that the best way out of this would be really
>>>>>>> to add annotation, so that the author of the pipeline can decide.
>>>>>>>
>>>>>>> If that would be acceptable I think I can try to prepare some
>>>>>>> basic functionality, but I'm not sure, if I would be able to
>>>>>>> cover all runners / sdks.
>>>>>>>
>>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>>>>>>>
>>>>>>> It is read all per key and window and not just read all (this
>>>>>>> still won't scale with hot keys in the global window). The GBK
>>>>>>> preceding the StatefulParDo will guarantee that you are
>>>>>>> processing all the values for a specific key and window at any
>>>>>>> given time. Is there a specific window/trigger that is missing
>>>>>>> that you feel would remove the need for you to use StatefulParDo?
>>>>>>>
>>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
>>>>>>> wrote:
>>>>>>>> Hi Lukasz,
>>>>>>>>
>>>>>>>>> Today, if you must have a strict order, you must guarantee
>>>>>>>>> that your StatefulParDo implements the necessary "buffering &
>>>>>>>>> sorting" into state.
>>>>>>>> Yes, no problem with that. But this whole discussion started,
>>>>>>>> because *this doesn't work on batch*. You simply cannot first
>>>>>>>> read everything from distributed storage and then buffer it all
>>>>>>>> into memory, just to read it again, but sorted. That will not
>>>>>>>> work. And even if it would, it would be a terrible waste of
>>>>>>>> resources.
>>>>>>>>
>>>>>>>> Jan
>>>>>>>>
>>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
>>>>>>>> wrote:
>>>>>>>>> This discussion brings many really interesting questions for
>>>>>>>>> me. :-)
>>>>>>>>>
>>>>>>>>> > I don't see batch vs. streaming as part of the model. One
>>>>>>>>> can have
>>>>>>>>> microbatch, or even a runner that alternates between different
>>>>>>>>> modes.
>>>>>>>>>
>>>>>>>>> Although I understand motivation of this statement, this
>>>>>>>>> project name is
>>>>>>>>> "Apache Beam: An advanced unified programming model". What
>>>>>>>>> does the
>>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>>>>>>>>>
>>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming
>>>>>>>>> are
>>>>>>>>> exactly the "runtime conditions/characteristics" I refer to.
>>>>>>>>> All these
>>>>>>>>> define several runtime parameters, which in turn define how
>>>>>>>>> well/badly
>>>>>>>>> will the pipeline perform and how many resources might be
>>>>>>>>> needed. From
>>>>>>>>> my point of view, pure streaming should be the most resource
>>>>>>>>> demanding
>>>>>>>>> (if not, why bother with batch? why not run everything in
>>>>>>>>> streaming
>>>>>>>>> only? what will there remain to "unify"?).
>>>>>>>>>
>>>>>>>>> > Fortunately, for batch, only the state for a single key
>>>>>>>>> needs to be
>>>>>>>>> preserved at a time, rather than the state for all keys across
>>>>>>>>> the range
>>>>>>>>> of skew. Of course if you have few or hot keys, one can still
>>>>>>>>> have
>>>>>>>>> issues (and this is not specific to StatefulDoFns).
>>>>>>>>>
>>>>>>>>> Yes, but here is still the presumption that my stateful DoFn can
>>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
>>>>>>>>> case in
>>>>>>>>> more detail.
>>>>>>>>>
>>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and
>>>>>>>>> some key for
>>>>>>>>> each element, which is irrelevant for the demonstration). Your
>>>>>>>>> task is
>>>>>>>>> to calculate in running global window the actual number of
>>>>>>>>> changes
>>>>>>>>> between state 0 and state 1 and vice versa. When the state
>>>>>>>>> doesn't
>>>>>>>>> change, you don't calculate anything. If input (for given key)
>>>>>>>>> would be
>>>>>>>>> (tN denotes timestamp N):
>>>>>>>>>
>>>>>>>>> t1: 1
>>>>>>>>>
>>>>>>>>> t2: 0
>>>>>>>>>
>>>>>>>>> t3: 0
>>>>>>>>>
>>>>>>>>> t4: 1
>>>>>>>>>
>>>>>>>>> t5: 1
>>>>>>>>>
>>>>>>>>> t6: 0
>>>>>>>>>
>>>>>>>>> then the output should yield (supposing that default state is
>>>>>>>>> zero):
>>>>>>>>>
>>>>>>>>> t1: (one: 1, zero: 0)
>>>>>>>>>
>>>>>>>>> t2: (one: 1, zero: 1)
>>>>>>>>>
>>>>>>>>> t3: (one: 1, zero: 1)
>>>>>>>>>
>>>>>>>>> t4: (one: 2, zero: 1)
>>>>>>>>>
>>>>>>>>> t5: (one: 2, zero: 1)
>>>>>>>>>
>>>>>>>>> t6: (one: 2, zero: 2)
>>>>>>>>>
>>>>>>>>> How would you implement this in current Beam semantics?
>>>>>>>> I think your saying here that I know that my input is ordered
>>>>>>>> in a specific way and since I assume the order when writing my
>>>>>>>> pipeline I can perform this optimization. But there is nothing
>>>>>>>> preventing a runner from noticing that your processing in the
>>>>>>>> global window with a specific type of trigger and re-ordering
>>>>>>>> your inputs/processing to get better performance (since you
>>>>>>>> can't use an AfterWatermark trigger for your pipeline in
>>>>>>>> streaming for the GlobalWindow).
>>>>>>>>
>>>>>>>> Today, if you must have a strict order, you must guarantee that
>>>>>>>> your StatefulParDo implements the necessary "buffering &
>>>>>>>> sorting" into state. I can see why you would want an annotation
>>>>>>>> that says I must have timestamp ordered elements, since it
>>>>>>>> makes writing certain StatefulParDos much easier. StatefulParDo
>>>>>>>> is a low-level function, it really is the "here you go and do
>>>>>>>> whatever you need to but here be dragons" function while
>>>>>>>> windowing and triggering is meant to keep many people from
>>>>>>>> writing StatefulParDo in the first place.
>>>>>>>>
>>>>>>>>> > Pipelines that fail in the "worst case" batch scenario
>>>>>>>>> are likely to
>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>>>>>>>>> falls
>>>>>>>>> behind in streaming mode as well.
>>>>>>>>>
>>>>>>>>> But the worst case is defined by input of size (available
>>>>>>>>> resources +
>>>>>>>>> single byte) -> pipeline fail. Although it could have
>>>>>>>>> finished, given
>>>>>>>>> the right conditions.
>>>>>>>>>
>>>>>>>>> > This might be reasonable, implemented by default by
>>>>>>>>> buffering
>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>>>>>>>>> advances,
>>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
>>>>>>>>> reason
>>>>>>>>> about) code.
>>>>>>>>>
>>>>>>>>> Sure, the pipeline will be less efficient, because it would
>>>>>>>>> have to
>>>>>>>>> buffer and sort the inputs. But at least it will produce
>>>>>>>>> correct results
>>>>>>>>> in cases where updates to state are order-sensitive.
>>>>>>>>>
>>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
>>>>>>>>> (key, values):
>>>>>>>>> [(key, value) for value in values])?
>>>>>>>>>
>>>>>>>>> I'd say roughly yes, but difference would be in the trigger.
>>>>>>>>> The trigger
>>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
>>>>>>>>> element
>>>>>>>>> with lowest timestamp in the buffer. Although this could be
>>>>>>>>> somehow
>>>>>>>>> emulated by fixed trigger each X millis.
>>>>>>>>>
>>>>>>>>> > Or is the underlying desire just to be able to hint to
>>>>>>>>> the runner
>>>>>>>>> that the code may perform better (e.g. require less resources)
>>>>>>>>> as skew
>>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
>>>>>>>>>
>>>>>>>>> No, the sorting would have to be done in streaming case as
>>>>>>>>> well. That is
>>>>>>>>> an imperative of the unified model. I think it is possible to
>>>>>>>>> sort by
>>>>>>>>> timestamp only in batch case (and do it for *all* batch
>>>>>>>>> stateful pardos
>>>>>>>>> without annotation), or introduce annotation, but then make
>>>>>>>>> the same
>>>>>>>>> guarantees for streaming case as well.
>>>>>>>>>
>>>>>>>>> Jan
>>>>>>>>>
>>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský
>>>>>>>>>> <je...@seznam.cz> wrote:
>>>>>>>>>>> Hi Robert,
>>>>>>>>>>>
>>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
>>>>>>>>>>> guarantees
>>>>>>>>>>> of ordering are given in either mode, there is *implicit*
>>>>>>>>>>> ordering in
>>>>>>>>>>> streaming case that is due to nature of the processing - the
>>>>>>>>>>> difference
>>>>>>>>>>> between watermark and timestamp of elements flowing through
>>>>>>>>>>> the pipeline
>>>>>>>>>>> are generally low (too high difference leads to the
>>>>>>>>>>> overbuffering
>>>>>>>>>>> problem), but there is no such bound on batch.
>>>>>>>>>> Fortunately, for batch, only the state for a single key needs
>>>>>>>>>> to be
>>>>>>>>>> preserved at a time, rather than the state for all keys
>>>>>>>>>> across the
>>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
>>>>>>>>>> still
>>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>>>>>>>>>>
>>>>>>>>>>> As a result, I see a few possible solutions:
>>>>>>>>>>>
>>>>>>>>>>> - the best and most natural seems to be extension of
>>>>>>>>>>> the model, so
>>>>>>>>>>> that it defines batch as not only "streaming pipeline
>>>>>>>>>>> executed in batch
>>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>>>>>>>>>>> characteristics as
>>>>>>>>>>> in streaming case, executed in batch fashion", I really
>>>>>>>>>>> don't think that
>>>>>>>>>>> there are any conflicts with the current model, or that this
>>>>>>>>>>> could
>>>>>>>>>>> affect performance, because the required sorting (as pointed by
>>>>>>>>>>> Aljoscha) is very probably already done during translation
>>>>>>>>>>> of stateful
>>>>>>>>>>> pardos. Also note that this definition only affects user
>>>>>>>>>>> defined
>>>>>>>>>>> stateful pardos
>>>>>>>>>> I don't see batch vs. streaming as part of the model. One can
>>>>>>>>>> have
>>>>>>>>>> microbatch, or even a runner that alternates between
>>>>>>>>>> different modes.
>>>>>>>>>> The model describes what the valid outputs are given a
>>>>>>>>>> (sometimes
>>>>>>>>>> partial) set of inputs. It becomes really hard to define
>>>>>>>>>> things like
>>>>>>>>>> "as good runtime characteristics." Once you allow any
>>>>>>>>>> out-of-orderedness, it is not very feasible to try and define
>>>>>>>>>> (and
>>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>>>>>>>>>> out-of-orderedness.
>>>>>>>>>>
>>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
>>>>>>>>>> likely to
>>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>>>>>>>>>> falls
>>>>>>>>>> behind in streaming mode as well.
>>>>>>>>>>
>>>>>>>>>>> - another option would be to introduce annotation for
>>>>>>>>>>> DoFns (e.g.
>>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in
>>>>>>>>>>> the sorting
>>>>>>>>>>> in batch case - but - this extension would have to ensure
>>>>>>>>>>> the sorting in
>>>>>>>>>>> streaming mode also - it would require definition of allowed
>>>>>>>>>>> lateness,
>>>>>>>>>>> and triggger (essentially similar to window)
>>>>>>>>>> This might be reasonable, implemented by default by buffering
>>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>>>>>>>>>> advances, but would likely lead to inefficient (though
>>>>>>>>>> *maybe* easier
>>>>>>>>>> to reason about) code. Not sure about the semantics of
>>>>>>>>>> triggering
>>>>>>>>>> here, especially data-driven triggers. Would it be roughly
>>>>>>>>>> equivalent
>>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for
>>>>>>>>>> value in
>>>>>>>>>> values])?
>>>>>>>>>>
>>>>>>>>>> Or is the underlying desire just to be able to hint to the
>>>>>>>>>> runner that
>>>>>>>>>> the code may perform better (e.g. require less resources) as
>>>>>>>>>> skew is
>>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>>>>>>>>>>
>>>>>>>>>>> - last option would be to introduce these "higher order
>>>>>>>>>>> guarantees" in
>>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
>>>>>>>>>>> worst
>>>>>>>>>>> option to me
>>>>>>>>>>>
>>>>>>>>>>> I see the first two options quite equally good, although the
>>>>>>>>>>> letter one
>>>>>>>>>>> is probably more time consuming to implement. But it would
>>>>>>>>>>> bring
>>>>>>>>>>> additional feature to streaming case as well.
>>>>>>>>>>>
>>>>>>>>>>> Thanks for any thoughts.
>>>>>>>>>>>
>>>>>>>>>>> Jan
>>>>>>>>>>>
>>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>>>>>>>>>>>> <je...@seznam.cz> wrote:
>>>>>>>>>>>>> Hi Reuven,
>>>>>>>>>>>>>
>>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch
>>>>>>>>>>>>>> runners.
>>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
>>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness
>>>>>>>>>>>>> of elements. That basically (practically) can work only
>>>>>>>>>>>>> for cases, where the order of input elements doesn't
>>>>>>>>>>>>> matter. But, "state" can refer to "state machine", and any
>>>>>>>>>>>>> time you have a state machine involved, then the ordering
>>>>>>>>>>>>> of elements would matter.
>>>>>>>>>>>> No guarantees on order are provided in *either* streaming
>>>>>>>>>>>> or batch
>>>>>>>>>>>> mode by the model. However, it is the case that in order to
>>>>>>>>>>>> make
>>>>>>>>>>>> forward progress most streaming runners attempt to limit
>>>>>>>>>>>> the amount of
>>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
>>>>>>>>>>>> processing
>>>>>>>>>>>> time) to make forward progress, which in turn could help
>>>>>>>>>>>> cap the
>>>>>>>>>>>> amount of state that must be held concurrently, whereas a
>>>>>>>>>>>> batch runner
>>>>>>>>>>>> may not allow any state to be safely discarded until the whole
>>>>>>>>>>>> timeline from infinite past to infinite future has been
>>>>>>>>>>>> observed.
>>>>>>>>>>>>
>>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to
>>>>>>>>>>>> batch" in batch mode.
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>>>>>>>>>>>> <mx...@apache.org> wrote:
>>>>>>>>>>>>
>>>>>>>>>>>>>> batch semantics and streaming semantics differs only
>>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
>>>>>>>>>>>>>> batch and cannot on stream
>>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
>>>>>>>>>>>>> trigger. You
>>>>>>>>>>>>> could define additional triggers that do early firings.
>>>>>>>>>>>>> And you could
>>>>>>>>>>>>> even trigger the global window by advancing the watermark
>>>>>>>>>>>>> to +inf.
>>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
>>>>>>>>>>>> default
>>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this
>>>>>>>>>>>> is more
>>>>>>>>>>>> likely to be user error than an actual desire to have no
>>>>>>>>>>>> output until
>>>>>>>>>>>> drain. But it's semantically valid in the model.
Re: Definition of Unified model
Posted by Maximilian Michels <mx...@apache.org>.
> Someone from Flink might correct me if I'm wrong, but that's my current understanding.
In essence your description of how exactly-once works in Flink is
correct. The general assumption in Flink is that pipelines must be
deterministic and thus produce idempotent writes in the case of
failures. However, that doesn't mean Beam sinks can't guarantee a bit
more with what Flink has to offer.
Luke already mentioned the design discussions for @RequiresStableInput
which ensures idempotent writes for non-deterministic pipelines. This is
not part of the model but an optional Beam feature.
We recently implemented support for @RequiresStableInput in the Flink
Runner. Reuven mentioned the Flink checkpoint confirmation, which allows
us to buffer (and checkpoint) processed data and only emit it once a
Flink checkpoint has completed.
Cheers,
Max
On 21.05.19 16:49, Jan Lukavský wrote:
> Hi,
>
> > Actually, I think it is a larger (open) question whether exactly once
> is guaranteed by the model or whether runners are allowed to relax that.
> I would think, however, that sources correctly implemented should be
> idempotent when run atop an exactly once infrastructure such as Flink of
> Dataflow.
>
> I would assume, that the model basically inherits guarantees of
> underlying infrastructure. Because Flink does not work as you described
> (atomic commit of inputs, state and outputs), but rather a checkpoint
> mark is flowing through the DAG much like watermark and on failures
> operators are restored and data reprocessed, it (IMHO) implies, that you
> have exactly once everywhere in the DAG *but* sinks. That is because
> sinks cannot be restored to previous state, instead sinks are supposed
> to be idempotent in order for the exactly once to really work (or at
> least be able to commit outputs on checkpoint in sink). That implies
> that if you don't have sink that is able to commit outputs atomically on
> checkpoint, the pipeline execution should be deterministic upon retries,
> otherwise shadow writes from failed paths of the pipeline might appear.
>
> Someone from Flink might correct me if I'm wrong, but that's my current
> understanding.
>
> > Sounds like we should make this clearer.
>
> I meant that you are right that we must not in any thoughts we are
> having forget that streams are by definition out-of-order. That is
> property that we cannot change. But - that doesn't limit us from
> creating operator that presents the data to UDF as if the stream was
> ideally sorted. It can do that by introducing latency, of course.
>
> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
>> Reza: One could provide something like this as a utility class, but
>> one downside is that it is not scale invariant. It requires a tuning
>> parameter that, if to small, won't mitigate the problem, but if to
>> big, greatly increases latency. (Possibly one could define a dynamic
>> session-like window to solve this though...) It also might be harder
>> for runners that *can* cheaply present stuff in timestamp order to
>> optimize. (That and, in practice, our annotation-style process methods
>> don't lend themselves to easy composition.) I think it could work in
>> specific cases though.
>>
>> More inline below.
>>
>> On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz> wrote:
>>> Hi Robert,
>>>
>>> > Beam has an exactly-once model. If the data was consumed, state
>>> mutated, and outputs written downstream (these three are committed
>>> together atomically) it will not be replayed. That does not, of course,
>>> solve the non-determanism due to ordering (including the fact that two
>>> operations reading the same PCollection may view different ordering).
>>>
>>> I think what you describe is a property of a runner, not of the model,
>>> right? I think if I run my pipeline on Flink I will not get this
>>> atomicity, because although Flink uses also exactly-once model if might
>>> write outputs multiple times.
>> Actually, I think it is a larger (open) question whether exactly once
>> is guaranteed by the model or whether runners are allowed to relax
>> that. I would think, however, that sources correctly implemented
>> should be idempotent when run atop an exactly once infrastructure such
>> as Flink of Dataflow.
>>
>>> > 1) Is it correct for a (Stateful)DoFn to assume elements are
>>> received
>>> in a specific order? In the current model, it is not. Being able to
>>> read, handle, and produced out-of-order data, including late data, is a
>>> pretty fundamental property of distributed systems.
>>>
>>> Yes, absolutely. The argument here is not that Stateful ParDo should
>>> presume to receive elements in any order, but to _present_ it as such to
>>> the user @ProcessElement function.
>> Sounds like we should make this clearer.
>>
>>> > 2) Given that some operations are easier (or possibly only possible)
>>> to write when operating on ordered data, and that different runners may
>>> have (significantly) cheaper ways to provide this ordering than can be
>>> done by the user themselves, should we elevate this to a property of
>>> (Stateful?)DoFns that the runner can provide? I think a compelling
>>> argument can be made here that we should.
>>>
>>> +1
>>>
>>> Jan
>>>
>>> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
>>>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz> wrote:
>>>>> > I don't see batch vs. streaming as part of the model. One can
>>>>> have
>>>>> microbatch, or even a runner that alternates between different modes.
>>>>>
>>>>> Although I understand motivation of this statement, this project
>>>>> name is
>>>>> "Apache Beam: An advanced unified programming model". What does the
>>>>> model unify, if "streaming vs. batch" is not part of the model?
>>>> What I mean is that streaming vs. batch is no longer part of the model
>>>> (or ideally API), but pushed down to be a concern of the runner
>>>> (executor) of the pipeline.
>>>>
>>>>
>>>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz> wrote:
>>>>> Hi Kenn,
>>>>>
>>>>> OK, so if we introduce annotation, we can have stateful ParDo with
>>>>> sorting, that would perfectly resolve my issues. I still have some
>>>>> doubts, though. Let me explain. The current behavior of stateful
>>>>> ParDo has the following properties:
>>>>>
>>>>> a) might fail in batch, although runs fine in streaming (that is
>>>>> due to the buffering, and unbounded lateness in batch, which was
>>>>> discussed back and forth in this thread)
>>>>>
>>>>> b) might be non deterministic (this is because the elements
>>>>> arrive at somewhat random order, and even if you do the operation
>>>>> "assign unique ID to elements" this might produce different results
>>>>> when run multiple times)
>>>> PCollections are *explicitly* unordered. Any operations that assume or
>>>> depend on a specific ordering for correctness (or determinism) must
>>>> provide that ordering themselves (i.e. tolerate "arbitrary shuffling
>>>> of inputs"). As you point out, that may be very expensive if you have
>>>> very hot keys with very large (unbounded) timestamp skew.
>>>>
>>>> StatefulDoFns are low-level operations that should be used with care;
>>>> the simpler windowing model gives determinism in the face of unordered
>>>> data (though late data and non-end-of-window triggering introduces
>>>> some of the non-determanism back in).
>>>>
>>>>> What worries me most is the property b), because it seems to me to
>>>>> have serious consequences - not only that if you run twice batch
>>>>> pipeline you would get different results, but even on streaming,
>>>>> when pipeline fails and gets restarted from checkpoint, produced
>>>>> output might differ from the previous run and data from the first
>>>>> run might have already been persisted into sink. That would create
>>>>> somewhat messy outputs.
>>>> Beam has an exactly-once model. If the data was consumed, state
>>>> mutated, and outputs written downstream (these three are committed
>>>> together atomically) it will not be replayed. That does not, of
>>>> course, solve the non-determanism due to ordering (including the fact
>>>> that two operations reading the same PCollection may view different
>>>> ordering).
>>>>
>>>>> These two properties makes me think that the current implementation
>>>>> is more of a _special case_ than the general one. The general one
>>>>> would be that your state doesn't have the properties to be able to
>>>>> tolerate buffering problems and/or non-determinism. Which is the
>>>>> case where you need sorting in both streaming and batch to be part
>>>>> of the model.
>>>>>
>>>>> Let me point out one more analogy - that is merging vs. non-merging
>>>>> windows. The general case (merging windows) implies sorting by
>>>>> timestamp in both batch case (explicit) and streaming (buffering).
>>>>> The special case (non-merging windows) doesn't rely on any
>>>>> timestamp ordering, so the sorting and buffering can be dropped.
>>>>> The underlying root cause of this is the same for both stateful
>>>>> ParDo and windowing (essentially, assigning window labels is a
>>>>> stateful operation when windowing function is merging).
>>>>>
>>>>> The reason for the current behavior of stateful ParDo seems to be
>>>>> performance, but is it right to abandon correctness in favor of
>>>>> performance? Wouldn't it be more consistent to have the default
>>>>> behavior prefer correctness and when you have the specific
>>>>> conditions of state function having special properties, then you
>>>>> can annotate your DoFn (with something like @TimeOrderingAgnostic),
>>>>> which would yield a better performance in that case?
>>>> There are two separable questions here.
>>>>
>>>> 1) Is it correct for a (Stateful)DoFn to assume elements are received
>>>> in a specific order? In the current model, it is not. Being able to
>>>> read, handle, and produced out-of-order data, including late data, is
>>>> a pretty fundamental property of distributed systems.
>>>>
>>>> 2) Given that some operations are easier (or possibly only possible)
>>>> to write when operating on ordered data, and that different runners
>>>> may have (significantly) cheaper ways to provide this ordering than
>>>> can be done by the user themselves, should we elevate this to a
>>>> property of (Stateful?)DoFns that the runner can provide? I think a
>>>> compelling argument can be made here that we should.
>>>>
>>>> - Robert
>>>>
>>>>
>>>>
>>>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
>>>>>
>>>>> Thanks for the nice small example of a calculation that depends on
>>>>> order. You are right that many state machines have this property. I
>>>>> agree w/ you and Luke that it is convenient for batch processing to
>>>>> sort by event timestamp before running a stateful ParDo. In
>>>>> streaming you could also implement "sort by event timestamp" by
>>>>> buffering until you know all earlier data will be dropped - a slack
>>>>> buffer up to allowed lateness.
>>>>>
>>>>> I do not think that it is OK to sort in batch and not in streaming.
>>>>> Many state machines diverge very rapidly when things are out of
>>>>> order. So each runner if they see the "@OrderByTimestamp"
>>>>> annotation (or whatever) needs to deliver sorted data (by some mix
>>>>> of buffering and dropping), or to reject the pipeline as unsupported.
>>>>>
>>>>> And also want to say that this is not the default case - many uses
>>>>> of state & timers in ParDo yield different results at the element
>>>>> level, but the results are equivalent at in the big picture. Such
>>>>> as the example of "assign a unique sequence number to each element"
>>>>> or "group into batches" it doesn't matter exactly what the result
>>>>> is, only that it meets the spec. And other cases like user funnels
>>>>> are monotonic enough that you also don't actually need sorting.
>>>>>
>>>>> Kenn
>>>>>
>>>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz> wrote:
>>>>>> Yes, the problem will arise probably mostly when you have not well
>>>>>> distributed keys (or too few keys). I'm really not sure if a pure
>>>>>> GBK with a trigger can solve this - it might help to have data
>>>>>> driven trigger. There would still be some doubts, though. The main
>>>>>> question is still here - people say, that sorting by timestamp
>>>>>> before stateful ParDo would be prohibitively slow, but I don't
>>>>>> really see why - the sorting is very probably already there. And
>>>>>> if not (hash grouping instead of sorted grouping), then the
>>>>>> sorting would affect only user defined StatefulParDos.
>>>>>>
>>>>>> This would suggest that the best way out of this would be really
>>>>>> to add annotation, so that the author of the pipeline can decide.
>>>>>>
>>>>>> If that would be acceptable I think I can try to prepare some
>>>>>> basic functionality, but I'm not sure, if I would be able to cover
>>>>>> all runners / sdks.
>>>>>>
>>>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
>>>>>>
>>>>>> It is read all per key and window and not just read all (this
>>>>>> still won't scale with hot keys in the global window). The GBK
>>>>>> preceding the StatefulParDo will guarantee that you are processing
>>>>>> all the values for a specific key and window at any given time. Is
>>>>>> there a specific window/trigger that is missing that you feel
>>>>>> would remove the need for you to use StatefulParDo?
>>>>>>
>>>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
>>>>>> wrote:
>>>>>>> Hi Lukasz,
>>>>>>>
>>>>>>>> Today, if you must have a strict order, you must guarantee that
>>>>>>>> your StatefulParDo implements the necessary "buffering &
>>>>>>>> sorting" into state.
>>>>>>> Yes, no problem with that. But this whole discussion started,
>>>>>>> because *this doesn't work on batch*. You simply cannot first
>>>>>>> read everything from distributed storage and then buffer it all
>>>>>>> into memory, just to read it again, but sorted. That will not
>>>>>>> work. And even if it would, it would be a terrible waste of
>>>>>>> resources.
>>>>>>>
>>>>>>> Jan
>>>>>>>
>>>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
>>>>>>> wrote:
>>>>>>>> This discussion brings many really interesting questions for me.
>>>>>>>> :-)
>>>>>>>>
>>>>>>>> > I don't see batch vs. streaming as part of the model. One
>>>>>>>> can have
>>>>>>>> microbatch, or even a runner that alternates between different
>>>>>>>> modes.
>>>>>>>>
>>>>>>>> Although I understand motivation of this statement, this project
>>>>>>>> name is
>>>>>>>> "Apache Beam: An advanced unified programming model". What does the
>>>>>>>> model unify, if "streaming vs. batch" is not part of the model?
>>>>>>>>
>>>>>>>> Using microbatching, chaining of batch jobs, or pure streaming are
>>>>>>>> exactly the "runtime conditions/characteristics" I refer to. All
>>>>>>>> these
>>>>>>>> define several runtime parameters, which in turn define how
>>>>>>>> well/badly
>>>>>>>> will the pipeline perform and how many resources might be
>>>>>>>> needed. From
>>>>>>>> my point of view, pure streaming should be the most resource
>>>>>>>> demanding
>>>>>>>> (if not, why bother with batch? why not run everything in streaming
>>>>>>>> only? what will there remain to "unify"?).
>>>>>>>>
>>>>>>>> > Fortunately, for batch, only the state for a single key
>>>>>>>> needs to be
>>>>>>>> preserved at a time, rather than the state for all keys across
>>>>>>>> the range
>>>>>>>> of skew. Of course if you have few or hot keys, one can still have
>>>>>>>> issues (and this is not specific to StatefulDoFns).
>>>>>>>>
>>>>>>>> Yes, but here is still the presumption that my stateful DoFn can
>>>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
>>>>>>>> case in
>>>>>>>> more detail.
>>>>>>>>
>>>>>>>> Suppose you have input stream consisting of 1s and 0s (and some
>>>>>>>> key for
>>>>>>>> each element, which is irrelevant for the demonstration). Your
>>>>>>>> task is
>>>>>>>> to calculate in running global window the actual number of changes
>>>>>>>> between state 0 and state 1 and vice versa. When the state doesn't
>>>>>>>> change, you don't calculate anything. If input (for given key)
>>>>>>>> would be
>>>>>>>> (tN denotes timestamp N):
>>>>>>>>
>>>>>>>> t1: 1
>>>>>>>>
>>>>>>>> t2: 0
>>>>>>>>
>>>>>>>> t3: 0
>>>>>>>>
>>>>>>>> t4: 1
>>>>>>>>
>>>>>>>> t5: 1
>>>>>>>>
>>>>>>>> t6: 0
>>>>>>>>
>>>>>>>> then the output should yield (supposing that default state is
>>>>>>>> zero):
>>>>>>>>
>>>>>>>> t1: (one: 1, zero: 0)
>>>>>>>>
>>>>>>>> t2: (one: 1, zero: 1)
>>>>>>>>
>>>>>>>> t3: (one: 1, zero: 1)
>>>>>>>>
>>>>>>>> t4: (one: 2, zero: 1)
>>>>>>>>
>>>>>>>> t5: (one: 2, zero: 1)
>>>>>>>>
>>>>>>>> t6: (one: 2, zero: 2)
>>>>>>>>
>>>>>>>> How would you implement this in current Beam semantics?
>>>>>>> I think your saying here that I know that my input is ordered in
>>>>>>> a specific way and since I assume the order when writing my
>>>>>>> pipeline I can perform this optimization. But there is nothing
>>>>>>> preventing a runner from noticing that your processing in the
>>>>>>> global window with a specific type of trigger and re-ordering
>>>>>>> your inputs/processing to get better performance (since you can't
>>>>>>> use an AfterWatermark trigger for your pipeline in streaming for
>>>>>>> the GlobalWindow).
>>>>>>>
>>>>>>> Today, if you must have a strict order, you must guarantee that
>>>>>>> your StatefulParDo implements the necessary "buffering & sorting"
>>>>>>> into state. I can see why you would want an annotation that says
>>>>>>> I must have timestamp ordered elements, since it makes writing
>>>>>>> certain StatefulParDos much easier. StatefulParDo is a low-level
>>>>>>> function, it really is the "here you go and do whatever you need
>>>>>>> to but here be dragons" function while windowing and triggering
>>>>>>> is meant to keep many people from writing StatefulParDo in the
>>>>>>> first place.
>>>>>>>
>>>>>>>> > Pipelines that fail in the "worst case" batch scenario are
>>>>>>>> likely to
>>>>>>>> degrade poorly (possibly catastrophically) when the watermark falls
>>>>>>>> behind in streaming mode as well.
>>>>>>>>
>>>>>>>> But the worst case is defined by input of size (available
>>>>>>>> resources +
>>>>>>>> single byte) -> pipeline fail. Although it could have finished,
>>>>>>>> given
>>>>>>>> the right conditions.
>>>>>>>>
>>>>>>>> > This might be reasonable, implemented by default by buffering
>>>>>>>> everything and releasing elements as the watermark (+lateness)
>>>>>>>> advances,
>>>>>>>> but would likely lead to inefficient (though *maybe* easier to
>>>>>>>> reason
>>>>>>>> about) code.
>>>>>>>>
>>>>>>>> Sure, the pipeline will be less efficient, because it would have to
>>>>>>>> buffer and sort the inputs. But at least it will produce correct
>>>>>>>> results
>>>>>>>> in cases where updates to state are order-sensitive.
>>>>>>>>
>>>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda
>>>>>>>> (key, values):
>>>>>>>> [(key, value) for value in values])?
>>>>>>>>
>>>>>>>> I'd say roughly yes, but difference would be in the trigger. The
>>>>>>>> trigger
>>>>>>>> should ideally fire as soon as watermark (+lateness) crosses
>>>>>>>> element
>>>>>>>> with lowest timestamp in the buffer. Although this could be somehow
>>>>>>>> emulated by fixed trigger each X millis.
>>>>>>>>
>>>>>>>> > Or is the underlying desire just to be able to hint to the
>>>>>>>> runner
>>>>>>>> that the code may perform better (e.g. require less resources)
>>>>>>>> as skew
>>>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
>>>>>>>>
>>>>>>>> No, the sorting would have to be done in streaming case as well.
>>>>>>>> That is
>>>>>>>> an imperative of the unified model. I think it is possible to
>>>>>>>> sort by
>>>>>>>> timestamp only in batch case (and do it for *all* batch stateful
>>>>>>>> pardos
>>>>>>>> without annotation), or introduce annotation, but then make the
>>>>>>>> same
>>>>>>>> guarantees for streaming case as well.
>>>>>>>>
>>>>>>>> Jan
>>>>>>>>
>>>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
>>>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský <je...@seznam.cz>
>>>>>>>>> wrote:
>>>>>>>>>> Hi Robert,
>>>>>>>>>>
>>>>>>>>>> yes, I think you rephrased my point - although no *explicit*
>>>>>>>>>> guarantees
>>>>>>>>>> of ordering are given in either mode, there is *implicit*
>>>>>>>>>> ordering in
>>>>>>>>>> streaming case that is due to nature of the processing - the
>>>>>>>>>> difference
>>>>>>>>>> between watermark and timestamp of elements flowing through
>>>>>>>>>> the pipeline
>>>>>>>>>> are generally low (too high difference leads to the overbuffering
>>>>>>>>>> problem), but there is no such bound on batch.
>>>>>>>>> Fortunately, for batch, only the state for a single key needs
>>>>>>>>> to be
>>>>>>>>> preserved at a time, rather than the state for all keys across the
>>>>>>>>> range of skew. Of course if you have few or hot keys, one can
>>>>>>>>> still
>>>>>>>>> have issues (and this is not specific to StatefulDoFns).
>>>>>>>>>
>>>>>>>>>> As a result, I see a few possible solutions:
>>>>>>>>>>
>>>>>>>>>> - the best and most natural seems to be extension of the
>>>>>>>>>> model, so
>>>>>>>>>> that it defines batch as not only "streaming pipeline executed
>>>>>>>>>> in batch
>>>>>>>>>> fashion", but "pipeline with at least as good runtime
>>>>>>>>>> characteristics as
>>>>>>>>>> in streaming case, executed in batch fashion", I really don't
>>>>>>>>>> think that
>>>>>>>>>> there are any conflicts with the current model, or that this
>>>>>>>>>> could
>>>>>>>>>> affect performance, because the required sorting (as pointed by
>>>>>>>>>> Aljoscha) is very probably already done during translation of
>>>>>>>>>> stateful
>>>>>>>>>> pardos. Also note that this definition only affects user defined
>>>>>>>>>> stateful pardos
>>>>>>>>> I don't see batch vs. streaming as part of the model. One can have
>>>>>>>>> microbatch, or even a runner that alternates between different
>>>>>>>>> modes.
>>>>>>>>> The model describes what the valid outputs are given a (sometimes
>>>>>>>>> partial) set of inputs. It becomes really hard to define things
>>>>>>>>> like
>>>>>>>>> "as good runtime characteristics." Once you allow any
>>>>>>>>> out-of-orderedness, it is not very feasible to try and define (and
>>>>>>>>> more cheaply implement) a "upper bound" of acceptable
>>>>>>>>> out-of-orderedness.
>>>>>>>>>
>>>>>>>>> Pipelines that fail in the "worst case" batch scenario are
>>>>>>>>> likely to
>>>>>>>>> degrade poorly (possibly catastrophically) when the watermark
>>>>>>>>> falls
>>>>>>>>> behind in streaming mode as well.
>>>>>>>>>
>>>>>>>>>> - another option would be to introduce annotation for
>>>>>>>>>> DoFns (e.g.
>>>>>>>>>> @RequiresStableTimeCharacteristics), which would result in the
>>>>>>>>>> sorting
>>>>>>>>>> in batch case - but - this extension would have to ensure the
>>>>>>>>>> sorting in
>>>>>>>>>> streaming mode also - it would require definition of allowed
>>>>>>>>>> lateness,
>>>>>>>>>> and triggger (essentially similar to window)
>>>>>>>>> This might be reasonable, implemented by default by buffering
>>>>>>>>> everything and releasing elements as the watermark (+lateness)
>>>>>>>>> advances, but would likely lead to inefficient (though *maybe*
>>>>>>>>> easier
>>>>>>>>> to reason about) code. Not sure about the semantics of triggering
>>>>>>>>> here, especially data-driven triggers. Would it be roughly
>>>>>>>>> equivalent
>>>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for value in
>>>>>>>>> values])?
>>>>>>>>>
>>>>>>>>> Or is the underlying desire just to be able to hint to the
>>>>>>>>> runner that
>>>>>>>>> the code may perform better (e.g. require less resources) as
>>>>>>>>> skew is
>>>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
>>>>>>>>>
>>>>>>>>>> - last option would be to introduce these "higher order
>>>>>>>>>> guarantees" in
>>>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
>>>>>>>>>> worst
>>>>>>>>>> option to me
>>>>>>>>>>
>>>>>>>>>> I see the first two options quite equally good, although the
>>>>>>>>>> letter one
>>>>>>>>>> is probably more time consuming to implement. But it would bring
>>>>>>>>>> additional feature to streaming case as well.
>>>>>>>>>>
>>>>>>>>>> Thanks for any thoughts.
>>>>>>>>>>
>>>>>>>>>> Jan
>>>>>>>>>>
>>>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
>>>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský
>>>>>>>>>>> <je...@seznam.cz> wrote:
>>>>>>>>>>>> Hi Reuven,
>>>>>>>>>>>>
>>>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch runners.
>>>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside
>>>>>>>>>>>> the state works for absolutely unbounded out-of-orderness of
>>>>>>>>>>>> elements. That basically (practically) can work only for
>>>>>>>>>>>> cases, where the order of input elements doesn't matter.
>>>>>>>>>>>> But, "state" can refer to "state machine", and any time you
>>>>>>>>>>>> have a state machine involved, then the ordering of elements
>>>>>>>>>>>> would matter.
>>>>>>>>>>> No guarantees on order are provided in *either* streaming or
>>>>>>>>>>> batch
>>>>>>>>>>> mode by the model. However, it is the case that in order to make
>>>>>>>>>>> forward progress most streaming runners attempt to limit the
>>>>>>>>>>> amount of
>>>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
>>>>>>>>>>> processing
>>>>>>>>>>> time) to make forward progress, which in turn could help cap the
>>>>>>>>>>> amount of state that must be held concurrently, whereas a
>>>>>>>>>>> batch runner
>>>>>>>>>>> may not allow any state to be safely discarded until the whole
>>>>>>>>>>> timeline from infinite past to infinite future has been
>>>>>>>>>>> observed.
>>>>>>>>>>>
>>>>>>>>>>> Also, as pointed out, state is not preserved "batch to batch"
>>>>>>>>>>> in batch mode.
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels
>>>>>>>>>>> <mx...@apache.org> wrote:
>>>>>>>>>>>
>>>>>>>>>>>>> batch semantics and streaming semantics differs only
>>>>>>>>>>>>> in that I can have GlobalWindow with default trigger on
>>>>>>>>>>>>> batch and cannot on stream
>>>>>>>>>>>> You can have a GlobalWindow in streaming with a default
>>>>>>>>>>>> trigger. You
>>>>>>>>>>>> could define additional triggers that do early firings. And
>>>>>>>>>>>> you could
>>>>>>>>>>>> even trigger the global window by advancing the watermark to
>>>>>>>>>>>> +inf.
>>>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
>>>>>>>>>>> default
>>>>>>>>>>> trigger on unbounded PCollections in the SDK because this is
>>>>>>>>>>> more
>>>>>>>>>>> likely to be user error than an actual desire to have no
>>>>>>>>>>> output until
>>>>>>>>>>> drain. But it's semantically valid in the model.
Re: Definition of Unified model
Posted by Reuven Lax <re...@google.com>.
I don't think this is completely accurate w.r.t Flink sinks. Flink provides
a way for sinks to buffer data until a snapshot has been performed, at
which point the data going to the sink is persistent. This has the exact
same effect as other runners (e.g. Dataflow) that persistently buffer data.
Beam uses this functionality - if a ParDo is tagged
with @RequiresStableInput.
As to whether the Beam model requires exactly once, that's a very good
question! In the past it was asserted that exactly once is not part of the
model (and in fact our compatibility matrix includes exactly once as a
feature). I think that this is the most "inclusive" definition, as it opens
the door for runners that can't support this. This does create some
interesting questions about what is meant by "model." In the mathematical
sense the model for a count operator that is not exactly once is that it
can return any value in [true_sum, infinity), which is not the most useful
of models.
On Tue, May 21, 2019 at 7:49 AM Jan Lukavský <je...@seznam.cz> wrote:
> Hi,
>
> > Actually, I think it is a larger (open) question whether exactly once
> is guaranteed by the model or whether runners are allowed to relax that.
> I would think, however, that sources correctly implemented should be
> idempotent when run atop an exactly once infrastructure such as Flink of
> Dataflow.
>
> I would assume, that the model basically inherits guarantees of
> underlying infrastructure. Because Flink does not work as you described
> (atomic commit of inputs, state and outputs), but rather a checkpoint
> mark is flowing through the DAG much like watermark and on failures
> operators are restored and data reprocessed, it (IMHO) implies, that you
> have exactly once everywhere in the DAG *but* sinks. That is because
> sinks cannot be restored to previous state, instead sinks are supposed
> to be idempotent in order for the exactly once to really work (or at
> least be able to commit outputs on checkpoint in sink). That implies
> that if you don't have sink that is able to commit outputs atomically on
> checkpoint, the pipeline execution should be deterministic upon retries,
> otherwise shadow writes from failed paths of the pipeline might appear.
>
> Someone from Flink might correct me if I'm wrong, but that's my current
> understanding.
>
> > Sounds like we should make this clearer.
>
> I meant that you are right that we must not in any thoughts we are
> having forget that streams are by definition out-of-order. That is
> property that we cannot change. But - that doesn't limit us from
> creating operator that presents the data to UDF as if the stream was
> ideally sorted. It can do that by introducing latency, of course.
>
> On 5/21/19 4:01 PM, Robert Bradshaw wrote:
> > Reza: One could provide something like this as a utility class, but
> > one downside is that it is not scale invariant. It requires a tuning
> > parameter that, if to small, won't mitigate the problem, but if to
> > big, greatly increases latency. (Possibly one could define a dynamic
> > session-like window to solve this though...) It also might be harder
> > for runners that *can* cheaply present stuff in timestamp order to
> > optimize. (That and, in practice, our annotation-style process methods
> > don't lend themselves to easy composition.) I think it could work in
> > specific cases though.
> >
> > More inline below.
> >
> > On Tue, May 21, 2019 at 11:38 AM Jan Lukavský <je...@seznam.cz> wrote:
> >> Hi Robert,
> >>
> >> > Beam has an exactly-once model. If the data was consumed, state
> >> mutated, and outputs written downstream (these three are committed
> >> together atomically) it will not be replayed. That does not, of course,
> >> solve the non-determanism due to ordering (including the fact that two
> >> operations reading the same PCollection may view different ordering).
> >>
> >> I think what you describe is a property of a runner, not of the model,
> >> right? I think if I run my pipeline on Flink I will not get this
> >> atomicity, because although Flink uses also exactly-once model if might
> >> write outputs multiple times.
> > Actually, I think it is a larger (open) question whether exactly once
> > is guaranteed by the model or whether runners are allowed to relax
> > that. I would think, however, that sources correctly implemented
> > should be idempotent when run atop an exactly once infrastructure such
> > as Flink of Dataflow.
> >
> >> > 1) Is it correct for a (Stateful)DoFn to assume elements are
> received
> >> in a specific order? In the current model, it is not. Being able to
> >> read, handle, and produced out-of-order data, including late data, is a
> >> pretty fundamental property of distributed systems.
> >>
> >> Yes, absolutely. The argument here is not that Stateful ParDo should
> >> presume to receive elements in any order, but to _present_ it as such to
> >> the user @ProcessElement function.
> > Sounds like we should make this clearer.
> >
> >> > 2) Given that some operations are easier (or possibly only possible)
> >> to write when operating on ordered data, and that different runners may
> >> have (significantly) cheaper ways to provide this ordering than can be
> >> done by the user themselves, should we elevate this to a property of
> >> (Stateful?)DoFns that the runner can provide? I think a compelling
> >> argument can be made here that we should.
> >>
> >> +1
> >>
> >> Jan
> >>
> >> On 5/21/19 11:07 AM, Robert Bradshaw wrote:
> >>> On Mon, May 20, 2019 at 5:24 PM Jan Lukavský <je...@seznam.cz> wrote:
> >>>> > I don't see batch vs. streaming as part of the model. One can
> have
> >>>> microbatch, or even a runner that alternates between different modes.
> >>>>
> >>>> Although I understand motivation of this statement, this project name
> is
> >>>> "Apache Beam: An advanced unified programming model". What does the
> >>>> model unify, if "streaming vs. batch" is not part of the model?
> >>> What I mean is that streaming vs. batch is no longer part of the model
> >>> (or ideally API), but pushed down to be a concern of the runner
> >>> (executor) of the pipeline.
> >>>
> >>>
> >>> On Tue, May 21, 2019 at 10:32 AM Jan Lukavský <je...@seznam.cz> wrote:
> >>>> Hi Kenn,
> >>>>
> >>>> OK, so if we introduce annotation, we can have stateful ParDo with
> sorting, that would perfectly resolve my issues. I still have some doubts,
> though. Let me explain. The current behavior of stateful ParDo has the
> following properties:
> >>>>
> >>>> a) might fail in batch, although runs fine in streaming (that is
> due to the buffering, and unbounded lateness in batch, which was discussed
> back and forth in this thread)
> >>>>
> >>>> b) might be non deterministic (this is because the elements arrive
> at somewhat random order, and even if you do the operation "assign unique
> ID to elements" this might produce different results when run multiple
> times)
> >>> PCollections are *explicitly* unordered. Any operations that assume or
> >>> depend on a specific ordering for correctness (or determinism) must
> >>> provide that ordering themselves (i.e. tolerate "arbitrary shuffling
> >>> of inputs"). As you point out, that may be very expensive if you have
> >>> very hot keys with very large (unbounded) timestamp skew.
> >>>
> >>> StatefulDoFns are low-level operations that should be used with care;
> >>> the simpler windowing model gives determinism in the face of unordered
> >>> data (though late data and non-end-of-window triggering introduces
> >>> some of the non-determanism back in).
> >>>
> >>>> What worries me most is the property b), because it seems to me to
> have serious consequences - not only that if you run twice batch pipeline
> you would get different results, but even on streaming, when pipeline fails
> and gets restarted from checkpoint, produced output might differ from the
> previous run and data from the first run might have already been persisted
> into sink. That would create somewhat messy outputs.
> >>> Beam has an exactly-once model. If the data was consumed, state
> >>> mutated, and outputs written downstream (these three are committed
> >>> together atomically) it will not be replayed. That does not, of
> >>> course, solve the non-determanism due to ordering (including the fact
> >>> that two operations reading the same PCollection may view different
> >>> ordering).
> >>>
> >>>> These two properties makes me think that the current implementation
> is more of a _special case_ than the general one. The general one would be
> that your state doesn't have the properties to be able to tolerate
> buffering problems and/or non-determinism. Which is the case where you need
> sorting in both streaming and batch to be part of the model.
> >>>>
> >>>> Let me point out one more analogy - that is merging vs. non-merging
> windows. The general case (merging windows) implies sorting by timestamp in
> both batch case (explicit) and streaming (buffering). The special case
> (non-merging windows) doesn't rely on any timestamp ordering, so the
> sorting and buffering can be dropped. The underlying root cause of this is
> the same for both stateful ParDo and windowing (essentially, assigning
> window labels is a stateful operation when windowing function is merging).
> >>>>
> >>>> The reason for the current behavior of stateful ParDo seems to be
> performance, but is it right to abandon correctness in favor of
> performance? Wouldn't it be more consistent to have the default behavior
> prefer correctness and when you have the specific conditions of state
> function having special properties, then you can annotate your DoFn (with
> something like @TimeOrderingAgnostic), which would yield a better
> performance in that case?
> >>> There are two separable questions here.
> >>>
> >>> 1) Is it correct for a (Stateful)DoFn to assume elements are received
> >>> in a specific order? In the current model, it is not. Being able to
> >>> read, handle, and produced out-of-order data, including late data, is
> >>> a pretty fundamental property of distributed systems.
> >>>
> >>> 2) Given that some operations are easier (or possibly only possible)
> >>> to write when operating on ordered data, and that different runners
> >>> may have (significantly) cheaper ways to provide this ordering than
> >>> can be done by the user themselves, should we elevate this to a
> >>> property of (Stateful?)DoFns that the runner can provide? I think a
> >>> compelling argument can be made here that we should.
> >>>
> >>> - Robert
> >>>
> >>>
> >>>
> >>>> On 5/21/19 1:00 AM, Kenneth Knowles wrote:
> >>>>
> >>>> Thanks for the nice small example of a calculation that depends on
> order. You are right that many state machines have this property. I agree
> w/ you and Luke that it is convenient for batch processing to sort by event
> timestamp before running a stateful ParDo. In streaming you could also
> implement "sort by event timestamp" by buffering until you know all earlier
> data will be dropped - a slack buffer up to allowed lateness.
> >>>>
> >>>> I do not think that it is OK to sort in batch and not in streaming.
> Many state machines diverge very rapidly when things are out of order. So
> each runner if they see the "@OrderByTimestamp" annotation (or whatever)
> needs to deliver sorted data (by some mix of buffering and dropping), or to
> reject the pipeline as unsupported.
> >>>>
> >>>> And also want to say that this is not the default case - many uses of
> state & timers in ParDo yield different results at the element level, but
> the results are equivalent at in the big picture. Such as the example of
> "assign a unique sequence number to each element" or "group into batches"
> it doesn't matter exactly what the result is, only that it meets the spec.
> And other cases like user funnels are monotonic enough that you also don't
> actually need sorting.
> >>>>
> >>>> Kenn
> >>>>
> >>>> On Mon, May 20, 2019 at 2:59 PM Jan Lukavský <je...@seznam.cz> wrote:
> >>>>> Yes, the problem will arise probably mostly when you have not well
> distributed keys (or too few keys). I'm really not sure if a pure GBK with
> a trigger can solve this - it might help to have data driven trigger. There
> would still be some doubts, though. The main question is still here -
> people say, that sorting by timestamp before stateful ParDo would be
> prohibitively slow, but I don't really see why - the sorting is very
> probably already there. And if not (hash grouping instead of sorted
> grouping), then the sorting would affect only user defined StatefulParDos.
> >>>>>
> >>>>> This would suggest that the best way out of this would be really to
> add annotation, so that the author of the pipeline can decide.
> >>>>>
> >>>>> If that would be acceptable I think I can try to prepare some basic
> functionality, but I'm not sure, if I would be able to cover all runners /
> sdks.
> >>>>>
> >>>>> On 5/20/19 11:36 PM, Lukasz Cwik wrote:
> >>>>>
> >>>>> It is read all per key and window and not just read all (this still
> won't scale with hot keys in the global window). The GBK preceding the
> StatefulParDo will guarantee that you are processing all the values for a
> specific key and window at any given time. Is there a specific
> window/trigger that is missing that you feel would remove the need for you
> to use StatefulParDo?
> >>>>>
> >>>>> On Mon, May 20, 2019 at 12:54 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>> Hi Lukasz,
> >>>>>>
> >>>>>>> Today, if you must have a strict order, you must guarantee that
> your StatefulParDo implements the necessary "buffering & sorting" into
> state.
> >>>>>> Yes, no problem with that. But this whole discussion started,
> because *this doesn't work on batch*. You simply cannot first read
> everything from distributed storage and then buffer it all into memory,
> just to read it again, but sorted. That will not work. And even if it
> would, it would be a terrible waste of resources.
> >>>>>>
> >>>>>> Jan
> >>>>>>
> >>>>>> On 5/20/19 8:39 PM, Lukasz Cwik wrote:
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> On Mon, May 20, 2019 at 8:24 AM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>>> This discussion brings many really interesting questions for me.
> :-)
> >>>>>>>
> >>>>>>> > I don't see batch vs. streaming as part of the model. One can
> have
> >>>>>>> microbatch, or even a runner that alternates between different
> modes.
> >>>>>>>
> >>>>>>> Although I understand motivation of this statement, this project
> name is
> >>>>>>> "Apache Beam: An advanced unified programming model". What does the
> >>>>>>> model unify, if "streaming vs. batch" is not part of the model?
> >>>>>>>
> >>>>>>> Using microbatching, chaining of batch jobs, or pure streaming are
> >>>>>>> exactly the "runtime conditions/characteristics" I refer to. All
> these
> >>>>>>> define several runtime parameters, which in turn define how
> well/badly
> >>>>>>> will the pipeline perform and how many resources might be needed.
> From
> >>>>>>> my point of view, pure streaming should be the most resource
> demanding
> >>>>>>> (if not, why bother with batch? why not run everything in streaming
> >>>>>>> only? what will there remain to "unify"?).
> >>>>>>>
> >>>>>>> > Fortunately, for batch, only the state for a single key needs
> to be
> >>>>>>> preserved at a time, rather than the state for all keys across the
> range
> >>>>>>> of skew. Of course if you have few or hot keys, one can still have
> >>>>>>> issues (and this is not specific to StatefulDoFns).
> >>>>>>>
> >>>>>>> Yes, but here is still the presumption that my stateful DoFn can
> >>>>>>> tolerate arbitrary shuffling of inputs. Let me explain the use
> case in
> >>>>>>> more detail.
> >>>>>>>
> >>>>>>> Suppose you have input stream consisting of 1s and 0s (and some
> key for
> >>>>>>> each element, which is irrelevant for the demonstration). Your
> task is
> >>>>>>> to calculate in running global window the actual number of changes
> >>>>>>> between state 0 and state 1 and vice versa. When the state doesn't
> >>>>>>> change, you don't calculate anything. If input (for given key)
> would be
> >>>>>>> (tN denotes timestamp N):
> >>>>>>>
> >>>>>>> t1: 1
> >>>>>>>
> >>>>>>> t2: 0
> >>>>>>>
> >>>>>>> t3: 0
> >>>>>>>
> >>>>>>> t4: 1
> >>>>>>>
> >>>>>>> t5: 1
> >>>>>>>
> >>>>>>> t6: 0
> >>>>>>>
> >>>>>>> then the output should yield (supposing that default state is
> zero):
> >>>>>>>
> >>>>>>> t1: (one: 1, zero: 0)
> >>>>>>>
> >>>>>>> t2: (one: 1, zero: 1)
> >>>>>>>
> >>>>>>> t3: (one: 1, zero: 1)
> >>>>>>>
> >>>>>>> t4: (one: 2, zero: 1)
> >>>>>>>
> >>>>>>> t5: (one: 2, zero: 1)
> >>>>>>>
> >>>>>>> t6: (one: 2, zero: 2)
> >>>>>>>
> >>>>>>> How would you implement this in current Beam semantics?
> >>>>>> I think your saying here that I know that my input is ordered in a
> specific way and since I assume the order when writing my pipeline I can
> perform this optimization. But there is nothing preventing a runner from
> noticing that your processing in the global window with a specific type of
> trigger and re-ordering your inputs/processing to get better performance
> (since you can't use an AfterWatermark trigger for your pipeline in
> streaming for the GlobalWindow).
> >>>>>>
> >>>>>> Today, if you must have a strict order, you must guarantee that
> your StatefulParDo implements the necessary "buffering & sorting" into
> state. I can see why you would want an annotation that says I must have
> timestamp ordered elements, since it makes writing certain StatefulParDos
> much easier. StatefulParDo is a low-level function, it really is the "here
> you go and do whatever you need to but here be dragons" function while
> windowing and triggering is meant to keep many people from writing
> StatefulParDo in the first place.
> >>>>>>
> >>>>>>> > Pipelines that fail in the "worst case" batch scenario are
> likely to
> >>>>>>> degrade poorly (possibly catastrophically) when the watermark falls
> >>>>>>> behind in streaming mode as well.
> >>>>>>>
> >>>>>>> But the worst case is defined by input of size (available
> resources +
> >>>>>>> single byte) -> pipeline fail. Although it could have finished,
> given
> >>>>>>> the right conditions.
> >>>>>>>
> >>>>>>> > This might be reasonable, implemented by default by buffering
> >>>>>>> everything and releasing elements as the watermark (+lateness)
> advances,
> >>>>>>> but would likely lead to inefficient (though *maybe* easier to
> reason
> >>>>>>> about) code.
> >>>>>>>
> >>>>>>> Sure, the pipeline will be less efficient, because it would have to
> >>>>>>> buffer and sort the inputs. But at least it will produce correct
> results
> >>>>>>> in cases where updates to state are order-sensitive.
> >>>>>>>
> >>>>>>> > Would it be roughly equivalent to GBK + FlatMap(lambda (key,
> values):
> >>>>>>> [(key, value) for value in values])?
> >>>>>>>
> >>>>>>> I'd say roughly yes, but difference would be in the trigger. The
> trigger
> >>>>>>> should ideally fire as soon as watermark (+lateness) crosses
> element
> >>>>>>> with lowest timestamp in the buffer. Although this could be somehow
> >>>>>>> emulated by fixed trigger each X millis.
> >>>>>>>
> >>>>>>> > Or is the underlying desire just to be able to hint to the
> runner
> >>>>>>> that the code may perform better (e.g. require less resources) as
> skew
> >>>>>>> is reduced (and hence to order by timestamp iff it's cheap)?
> >>>>>>>
> >>>>>>> No, the sorting would have to be done in streaming case as well.
> That is
> >>>>>>> an imperative of the unified model. I think it is possible to sort
> by
> >>>>>>> timestamp only in batch case (and do it for *all* batch stateful
> pardos
> >>>>>>> without annotation), or introduce annotation, but then make the
> same
> >>>>>>> guarantees for streaming case as well.
> >>>>>>>
> >>>>>>> Jan
> >>>>>>>
> >>>>>>> On 5/20/19 4:41 PM, Robert Bradshaw wrote:
> >>>>>>>> On Mon, May 20, 2019 at 1:19 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>>>>> Hi Robert,
> >>>>>>>>>
> >>>>>>>>> yes, I think you rephrased my point - although no *explicit*
> guarantees
> >>>>>>>>> of ordering are given in either mode, there is *implicit*
> ordering in
> >>>>>>>>> streaming case that is due to nature of the processing - the
> difference
> >>>>>>>>> between watermark and timestamp of elements flowing through the
> pipeline
> >>>>>>>>> are generally low (too high difference leads to the overbuffering
> >>>>>>>>> problem), but there is no such bound on batch.
> >>>>>>>> Fortunately, for batch, only the state for a single key needs to
> be
> >>>>>>>> preserved at a time, rather than the state for all keys across the
> >>>>>>>> range of skew. Of course if you have few or hot keys, one can
> still
> >>>>>>>> have issues (and this is not specific to StatefulDoFns).
> >>>>>>>>
> >>>>>>>>> As a result, I see a few possible solutions:
> >>>>>>>>>
> >>>>>>>>> - the best and most natural seems to be extension of the
> model, so
> >>>>>>>>> that it defines batch as not only "streaming pipeline executed
> in batch
> >>>>>>>>> fashion", but "pipeline with at least as good runtime
> characteristics as
> >>>>>>>>> in streaming case, executed in batch fashion", I really don't
> think that
> >>>>>>>>> there are any conflicts with the current model, or that this
> could
> >>>>>>>>> affect performance, because the required sorting (as pointed by
> >>>>>>>>> Aljoscha) is very probably already done during translation of
> stateful
> >>>>>>>>> pardos. Also note that this definition only affects user defined
> >>>>>>>>> stateful pardos
> >>>>>>>> I don't see batch vs. streaming as part of the model. One can have
> >>>>>>>> microbatch, or even a runner that alternates between different
> modes.
> >>>>>>>> The model describes what the valid outputs are given a (sometimes
> >>>>>>>> partial) set of inputs. It becomes really hard to define things
> like
> >>>>>>>> "as good runtime characteristics." Once you allow any
> >>>>>>>> out-of-orderedness, it is not very feasible to try and define (and
> >>>>>>>> more cheaply implement) a "upper bound" of acceptable
> >>>>>>>> out-of-orderedness.
> >>>>>>>>
> >>>>>>>> Pipelines that fail in the "worst case" batch scenario are likely
> to
> >>>>>>>> degrade poorly (possibly catastrophically) when the watermark
> falls
> >>>>>>>> behind in streaming mode as well.
> >>>>>>>>
> >>>>>>>>> - another option would be to introduce annotation for DoFns
> (e.g.
> >>>>>>>>> @RequiresStableTimeCharacteristics), which would result in the
> sorting
> >>>>>>>>> in batch case - but - this extension would have to ensure the
> sorting in
> >>>>>>>>> streaming mode also - it would require definition of allowed
> lateness,
> >>>>>>>>> and triggger (essentially similar to window)
> >>>>>>>> This might be reasonable, implemented by default by buffering
> >>>>>>>> everything and releasing elements as the watermark (+lateness)
> >>>>>>>> advances, but would likely lead to inefficient (though *maybe*
> easier
> >>>>>>>> to reason about) code. Not sure about the semantics of triggering
> >>>>>>>> here, especially data-driven triggers. Would it be roughly
> equivalent
> >>>>>>>> to GBK + FlatMap(lambda (key, values): [(key, value) for value in
> >>>>>>>> values])?
> >>>>>>>>
> >>>>>>>> Or is the underlying desire just to be able to hint to the runner
> that
> >>>>>>>> the code may perform better (e.g. require less resources) as skew
> is
> >>>>>>>> reduced (and hence to order by timestamp iff it's cheap)?
> >>>>>>>>
> >>>>>>>>> - last option would be to introduce these "higher order
> guarantees" in
> >>>>>>>>> some extension DSL (e.g. Euphoria), but that seems to be the
> worst
> >>>>>>>>> option to me
> >>>>>>>>>
> >>>>>>>>> I see the first two options quite equally good, although the
> letter one
> >>>>>>>>> is probably more time consuming to implement. But it would bring
> >>>>>>>>> additional feature to streaming case as well.
> >>>>>>>>>
> >>>>>>>>> Thanks for any thoughts.
> >>>>>>>>>
> >>>>>>>>> Jan
> >>>>>>>>>
> >>>>>>>>> On 5/20/19 12:41 PM, Robert Bradshaw wrote:
> >>>>>>>>>> On Fri, May 17, 2019 at 4:48 PM Jan Lukavský <je...@seznam.cz>
> wrote:
> >>>>>>>>>>> Hi Reuven,
> >>>>>>>>>>>
> >>>>>>>>>>>> How so? AFAIK stateful DoFns work just fine in batch runners.
> >>>>>>>>>>> Stateful ParDo works in batch as far, as the logic inside the
> state works for absolutely unbounded out-of-orderness of elements. That
> basically (practically) can work only for cases, where the order of input
> elements doesn't matter. But, "state" can refer to "state machine", and any
> time you have a state machine involved, then the ordering of elements would
> matter.
> >>>>>>>>>> No guarantees on order are provided in *either* streaming or
> batch
> >>>>>>>>>> mode by the model. However, it is the case that in order to make
> >>>>>>>>>> forward progress most streaming runners attempt to limit the
> amount of
> >>>>>>>>>> out-of-orderedness of elements (in terms of event time vs.
> processing
> >>>>>>>>>> time) to make forward progress, which in turn could help cap the
> >>>>>>>>>> amount of state that must be held concurrently, whereas a batch
> runner
> >>>>>>>>>> may not allow any state to be safely discarded until the whole
> >>>>>>>>>> timeline from infinite past to infinite future has been
> observed.
> >>>>>>>>>>
> >>>>>>>>>> Also, as pointed out, state is not preserved "batch to batch"
> in batch mode.
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> On Thu, May 16, 2019 at 3:59 PM Maximilian Michels <
> mxm@apache.org> wrote:
> >>>>>>>>>>
> >>>>>>>>>>>> batch semantics and streaming semantics differs only in
> that I can have GlobalWindow with default trigger on batch and cannot on
> stream
> >>>>>>>>>>> You can have a GlobalWindow in streaming with a default
> trigger. You
> >>>>>>>>>>> could define additional triggers that do early firings. And
> you could
> >>>>>>>>>>> even trigger the global window by advancing the watermark to
> +inf.
> >>>>>>>>>> IIRC, as a pragmatic note, we prohibited global window with
> default
> >>>>>>>>>> trigger on unbounded PCollections in the SDK because this is
> more
> >>>>>>>>>> likely to be user error than an actual desire to have no output
> until
> >>>>>>>>>> drain. But it's semantically valid in the model.
>