You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@accumulo.apache.org by Ara Ebrahimi <ar...@argyledata.com> on 2015/02/19 06:57:30 UTC
OfflineScanner
Hi,
I’m trying to optimize a connector we’ve written for Presto. In some cases we need to perform full table scans. This happens across all the nodes but each node is assigned to process only a sharded subset of data. Each shard is hosted by only 1 RFile. I’m looking at the AbstractInputFormat and OfflineIterator and it seems like the code is not that hard to use for this case. Is there any drawback? It seems like if the table is offline then OfflineIterator is used which apparently reads the RFiles directly and doesn’t involve any RPC and I think should be significantly faster. Is it so? Is there any drawback to using this while the table is not offline but no other app is messing with the table?
Thanks,
Ara.
________________________________
This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
________________________________
Re: OfflineScanner
Posted by Sean Busbey <bu...@cloudera.com>.
Hi Marc!
Yep, you can do this using the optional "setOfflineTableScan" on
AccumuloInputFormat[1]. It still requires that the table be offline.
There's a good example of programmatically creating an offline clone if you
look at the MR job we use to verify the "Continuous Ingest" integration
test[2]:
----
Random random = new Random();
clone = opts.getTableName() + "_" + String.format("%016x",
(random.nextLong() & 0x7fffffffffffffffl));
conn = opts.getConnector();
conn.tableOperations().clone(opts.getTableName(), clone, true, new
HashMap<String,String>(), new HashSet<String>());
ranges =
conn.tableOperations().splitRangeByTablets(opts.getTableName(), new
Range(), opts.maxMaps);
conn.tableOperations().offline(clone);
AccumuloInputFormat.setInputTableName(job, clone);
AccumuloInputFormat.setOfflineTableScan(job, true);
----
[1]: *http://s.apache.org/Pul <http://s.apache.org/Pul>*
[2]: *http://s.apache.org/P6Z <http://s.apache.org/P6Z>*
On Thu, Feb 19, 2015 at 9:47 AM, Marc Reichman <mreichman@pixelforensics.com
> wrote:
> Apologies for hijacking this, but is there any way to use an offline table
> clone with MapReduce and AccumuloInputFormat? That read speed increase
> sounds very appealing..
>
> On Thu, Feb 19, 2015 at 9:27 AM, Josh Elser <jo...@gmail.com> wrote:
>
>> Typically, if you're using the OfflineScanner, you'd clone the table you
>> want to read and then take the clone offline. It's a simple (and fast)
>> solution that doesn't interrupt the availability of the table.
>>
>> Doing the read offline will definitely be faster (maybe 20%, I'm not
>> entirely sure on accurate number and how it scales with nodes). The pain
>> would be the extra work in creating the clone, offline'ing the table, and
>> eventually deleting the clone when you're done with it. A little more work,
>> but manageable.
>>
>>
>> Ara Ebrahimi wrote:
>>
>>> Hi,
>>>
>>> I’m trying to optimize a connector we’ve written for Presto. In some
>>> cases we need to perform full table scans. This happens across all the
>>> nodes but each node is assigned to process only a sharded subset of data.
>>> Each shard is hosted by only 1 RFile. I’m looking at the
>>> AbstractInputFormat and OfflineIterator and it seems like the code is not
>>> that hard to use for this case. Is there any drawback? It seems like if the
>>> table is offline then OfflineIterator is used which apparently reads the
>>> RFiles directly and doesn’t involve any RPC and I think should be
>>> significantly faster. Is it so? Is there any drawback to using this while
>>> the table is not offline but no other app is messing with the table?
>>>
>>> Thanks,
>>> Ara.
>>>
>>>
>>>
>>> ________________________________
>>>
>>> This message is for the designated recipient only and may contain
>>> privileged, proprietary, or otherwise confidential information. If you have
>>> received it in error, please notify the sender immediately and delete the
>>> original. Any other use of the e-mail by you is prohibited. Thank you in
>>> advance for your cooperation.
>>>
>>> ________________________________
>>>
>>
>
--
Sean
Re: OfflineScanner
Posted by Marc Reichman <mr...@pixelforensics.com>.
Apologies for hijacking this, but is there any way to use an offline table
clone with MapReduce and AccumuloInputFormat? That read speed increase
sounds very appealing..
On Thu, Feb 19, 2015 at 9:27 AM, Josh Elser <jo...@gmail.com> wrote:
> Typically, if you're using the OfflineScanner, you'd clone the table you
> want to read and then take the clone offline. It's a simple (and fast)
> solution that doesn't interrupt the availability of the table.
>
> Doing the read offline will definitely be faster (maybe 20%, I'm not
> entirely sure on accurate number and how it scales with nodes). The pain
> would be the extra work in creating the clone, offline'ing the table, and
> eventually deleting the clone when you're done with it. A little more work,
> but manageable.
>
>
> Ara Ebrahimi wrote:
>
>> Hi,
>>
>> I’m trying to optimize a connector we’ve written for Presto. In some
>> cases we need to perform full table scans. This happens across all the
>> nodes but each node is assigned to process only a sharded subset of data.
>> Each shard is hosted by only 1 RFile. I’m looking at the
>> AbstractInputFormat and OfflineIterator and it seems like the code is not
>> that hard to use for this case. Is there any drawback? It seems like if the
>> table is offline then OfflineIterator is used which apparently reads the
>> RFiles directly and doesn’t involve any RPC and I think should be
>> significantly faster. Is it so? Is there any drawback to using this while
>> the table is not offline but no other app is messing with the table?
>>
>> Thanks,
>> Ara.
>>
>>
>>
>> ________________________________
>>
>> This message is for the designated recipient only and may contain
>> privileged, proprietary, or otherwise confidential information. If you have
>> received it in error, please notify the sender immediately and delete the
>> original. Any other use of the e-mail by you is prohibited. Thank you in
>> advance for your cooperation.
>>
>> ________________________________
>>
>
Re: OfflineScanner
Posted by Ara Ebrahimi <ar...@argyledata.com>.
Ok so I need to clone it. Doable. Thanks!
Ara.
> On Feb 19, 2015, at 7:28 AM, Josh Elser <jo...@gmail.com> wrote:
>
> Typically, if you're using the OfflineScanner, you'd clone the table you
> want to read and then take the clone offline. It's a simple (and fast)
> solution that doesn't interrupt the availability of the table.
>
> Doing the read offline will definitely be faster (maybe 20%, I'm not
> entirely sure on accurate number and how it scales with nodes). The pain
> would be the extra work in creating the clone, offline'ing the table,
> and eventually deleting the clone when you're done with it. A little
> more work, but manageable.
>
> Ara Ebrahimi wrote:
>> Hi,
>>
>> I’m trying to optimize a connector we’ve written for Presto. In some cases we need to perform full table scans. This happens across all the nodes but each node is assigned to process only a sharded subset of data. Each shard is hosted by only 1 RFile. I’m looking at the AbstractInputFormat and OfflineIterator and it seems like the code is not that hard to use for this case. Is there any drawback? It seems like if the table is offline then OfflineIterator is used which apparently reads the RFiles directly and doesn’t involve any RPC and I think should be significantly faster. Is it so? Is there any drawback to using this while the table is not offline but no other app is messing with the table?
>>
>> Thanks,
>> Ara.
>>
>>
>>
>> ________________________________
>>
>> This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
>>
>> ________________________________
>
>
>
> ________________________________
>
> This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
>
> ________________________________
________________________________
This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
________________________________
Re: OfflineScanner
Posted by Josh Elser <jo...@gmail.com>.
Typically, if you're using the OfflineScanner, you'd clone the table you
want to read and then take the clone offline. It's a simple (and fast)
solution that doesn't interrupt the availability of the table.
Doing the read offline will definitely be faster (maybe 20%, I'm not
entirely sure on accurate number and how it scales with nodes). The pain
would be the extra work in creating the clone, offline'ing the table,
and eventually deleting the clone when you're done with it. A little
more work, but manageable.
Ara Ebrahimi wrote:
> Hi,
>
> I’m trying to optimize a connector we’ve written for Presto. In some cases we need to perform full table scans. This happens across all the nodes but each node is assigned to process only a sharded subset of data. Each shard is hosted by only 1 RFile. I’m looking at the AbstractInputFormat and OfflineIterator and it seems like the code is not that hard to use for this case. Is there any drawback? It seems like if the table is offline then OfflineIterator is used which apparently reads the RFiles directly and doesn’t involve any RPC and I think should be significantly faster. Is it so? Is there any drawback to using this while the table is not offline but no other app is messing with the table?
>
> Thanks,
> Ara.
>
>
>
> ________________________________
>
> This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
>
> ________________________________
Re: OfflineScanner
Posted by Josh Elser <jo...@gmail.com>.
Even so, as it stands, we have no "obligation" to make sure code you
write that uses OfflineScanner will work across versions. Granted, we're
not going to go and do it just to mess with you, but it's something that
we can do to help you sleep better at night.
Ara Ebrahimi wrote:
> Actually I was wrong. OfflineScanner is public, OfflineIterator is package. So it’s good enough :)
>
> Ara.
>
>> On Feb 19, 2015, at 11:46 AM, Josh Elser<jo...@gmail.com> wrote:
>>
>> Want to file a ticket, Ara? I didn't realize it wasn't directly in the
>> public API (only via m/r). I think it would make a nice addition.
>>
>> Ara Ebrahimi wrote:
>>> OfflineScanner is package protected. So I'll need to hack it. If it
>>> proves to be faster at least 20% then it's worth having it in the public
>>> Ali, perhaps even let user use it by a asking specific file to be
>>> scanned rather than directing scan by carefully defining the range to
>>> touch the intended file.
>>>
>>> Ara.
>>>
>>> On Feb 19, 2015, at 8:15 AM, Keith Turner<keith@deenlo.com
>>> <ma...@deenlo.com>> wrote:
>>>
>>>>
>>>> On Thu, Feb 19, 2015 at 12:57 AM, Ara Ebrahimi
>>>> <ar...@argyledata.com>> wrote:
>>>>
>>>> Hi,
>>>>
>>>> I’m trying to optimize a connector we’ve written for Presto. In
>>>> some cases we need to perform full table scans. This happens
>>>> across all the nodes but each node is assigned to process only a
>>>> sharded subset of data. Each shard is hosted by only 1 RFile. I’m
>>>> looking at the AbstractInputFormat and OfflineIterator and it
>>>> seems like the code is not that hard to use for this case. Is
>>>> there any drawback? It seems like if the table is offline then
>>>> OfflineIterator is used which apparently reads the RFiles directly
>>>> and doesn’t involve any RPC and I think should be significantly
>>>> faster. Is it so? Is there any drawback to using this while the
>>>> table is not offline but no other app is messing with the table?
>>>>
>>>>
>>>> The code will throw an exception if the table is not offline (intent
>>>> is to ensure the files are stable and not garbage collected). As
>>>> others have stated you can clone.
>>>> Currently offline scanning is only supported in the public API w/ Map
>>>> Reduce. Curious, would you be interested in seeing this in the client
>>>> public API?
>>>>
>>>>
>>>> Thanks,
>>>> Ara.
>>>>
>>>>
>>>>
>>>> ________________________________
>>>>
>>>> This message is for the designated recipient only and may contain
>>>> privileged, proprietary, or otherwise confidential information. If
>>>> you have received it in error, please notify the sender
>>>> immediately and delete the original. Any other use of the e-mail
>>>> by you is prohibited. Thank you in advance for your cooperation.
>>>>
>>>> ________________________________
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>> This message is for the designated recipient only and may contain
>>>> privileged, proprietary, or otherwise confidential information. If you
>>>> have received it in error, please notify the sender immediately and
>>>> delete the original. Any other use of the e-mail by you is prohibited.
>>>> Thank you in advance for your cooperation.
>>>>
>>>> ------------------------------------------------------------------------
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>> This message is for the designated recipient only and may contain
>>> privileged, proprietary, or otherwise confidential information. If you
>>> have received it in error, please notify the sender immediately and
>>> delete the original. Any other use of the e-mail by you is prohibited.
>>> Thank you in advance for your cooperation.
>>>
>>> ------------------------------------------------------------------------
>>
>>
>> ________________________________
>>
>> This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
>>
>> ________________________________
>
>
>
>
> ________________________________
>
> This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
>
> ________________________________
Re: OfflineScanner
Posted by Ara Ebrahimi <ar...@argyledata.com>.
Actually I was wrong. OfflineScanner is public, OfflineIterator is package. So it’s good enough :)
Ara.
> On Feb 19, 2015, at 11:46 AM, Josh Elser <jo...@gmail.com> wrote:
>
> Want to file a ticket, Ara? I didn't realize it wasn't directly in the
> public API (only via m/r). I think it would make a nice addition.
>
> Ara Ebrahimi wrote:
>> OfflineScanner is package protected. So I'll need to hack it. If it
>> proves to be faster at least 20% then it's worth having it in the public
>> Ali, perhaps even let user use it by a asking specific file to be
>> scanned rather than directing scan by carefully defining the range to
>> touch the intended file.
>>
>> Ara.
>>
>> On Feb 19, 2015, at 8:15 AM, Keith Turner <keith@deenlo.com
>> <ma...@deenlo.com>> wrote:
>>
>>>
>>>
>>> On Thu, Feb 19, 2015 at 12:57 AM, Ara Ebrahimi
>>> <ara.ebrahimi@argyledata.com <ma...@argyledata.com>> wrote:
>>>
>>> Hi,
>>>
>>> I’m trying to optimize a connector we’ve written for Presto. In
>>> some cases we need to perform full table scans. This happens
>>> across all the nodes but each node is assigned to process only a
>>> sharded subset of data. Each shard is hosted by only 1 RFile. I’m
>>> looking at the AbstractInputFormat and OfflineIterator and it
>>> seems like the code is not that hard to use for this case. Is
>>> there any drawback? It seems like if the table is offline then
>>> OfflineIterator is used which apparently reads the RFiles directly
>>> and doesn’t involve any RPC and I think should be significantly
>>> faster. Is it so? Is there any drawback to using this while the
>>> table is not offline but no other app is messing with the table?
>>>
>>>
>>> The code will throw an exception if the table is not offline (intent
>>> is to ensure the files are stable and not garbage collected). As
>>> others have stated you can clone.
>>> Currently offline scanning is only supported in the public API w/ Map
>>> Reduce. Curious, would you be interested in seeing this in the client
>>> public API?
>>>
>>>
>>> Thanks,
>>> Ara.
>>>
>>>
>>>
>>> ________________________________
>>>
>>> This message is for the designated recipient only and may contain
>>> privileged, proprietary, or otherwise confidential information. If
>>> you have received it in error, please notify the sender
>>> immediately and delete the original. Any other use of the e-mail
>>> by you is prohibited. Thank you in advance for your cooperation.
>>>
>>> ________________________________
>>>
>>>
>>>
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>> This message is for the designated recipient only and may contain
>>> privileged, proprietary, or otherwise confidential information. If you
>>> have received it in error, please notify the sender immediately and
>>> delete the original. Any other use of the e-mail by you is prohibited.
>>> Thank you in advance for your cooperation.
>>>
>>> ------------------------------------------------------------------------
>>
>>
>>
>> ------------------------------------------------------------------------
>>
>> This message is for the designated recipient only and may contain
>> privileged, proprietary, or otherwise confidential information. If you
>> have received it in error, please notify the sender immediately and
>> delete the original. Any other use of the e-mail by you is prohibited.
>> Thank you in advance for your cooperation.
>>
>> ------------------------------------------------------------------------
>
>
>
> ________________________________
>
> This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
>
> ________________________________
________________________________
This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
________________________________
Re: OfflineScanner
Posted by Josh Elser <jo...@gmail.com>.
Want to file a ticket, Ara? I didn't realize it wasn't directly in the
public API (only via m/r). I think it would make a nice addition.
Ara Ebrahimi wrote:
> OfflineScanner is package protected. So I'll need to hack it. If it
> proves to be faster at least 20% then it's worth having it in the public
> Ali, perhaps even let user use it by a asking specific file to be
> scanned rather than directing scan by carefully defining the range to
> touch the intended file.
>
> Ara.
>
> On Feb 19, 2015, at 8:15 AM, Keith Turner <keith@deenlo.com
> <ma...@deenlo.com>> wrote:
>
>>
>>
>> On Thu, Feb 19, 2015 at 12:57 AM, Ara Ebrahimi
>> <ara.ebrahimi@argyledata.com <ma...@argyledata.com>> wrote:
>>
>> Hi,
>>
>> I’m trying to optimize a connector we’ve written for Presto. In
>> some cases we need to perform full table scans. This happens
>> across all the nodes but each node is assigned to process only a
>> sharded subset of data. Each shard is hosted by only 1 RFile. I’m
>> looking at the AbstractInputFormat and OfflineIterator and it
>> seems like the code is not that hard to use for this case. Is
>> there any drawback? It seems like if the table is offline then
>> OfflineIterator is used which apparently reads the RFiles directly
>> and doesn’t involve any RPC and I think should be significantly
>> faster. Is it so? Is there any drawback to using this while the
>> table is not offline but no other app is messing with the table?
>>
>>
>> The code will throw an exception if the table is not offline (intent
>> is to ensure the files are stable and not garbage collected). As
>> others have stated you can clone.
>> Currently offline scanning is only supported in the public API w/ Map
>> Reduce. Curious, would you be interested in seeing this in the client
>> public API?
>>
>>
>> Thanks,
>> Ara.
>>
>>
>>
>> ________________________________
>>
>> This message is for the designated recipient only and may contain
>> privileged, proprietary, or otherwise confidential information. If
>> you have received it in error, please notify the sender
>> immediately and delete the original. Any other use of the e-mail
>> by you is prohibited. Thank you in advance for your cooperation.
>>
>> ________________________________
>>
>>
>>
>>
>>
>> ------------------------------------------------------------------------
>>
>> This message is for the designated recipient only and may contain
>> privileged, proprietary, or otherwise confidential information. If you
>> have received it in error, please notify the sender immediately and
>> delete the original. Any other use of the e-mail by you is prohibited.
>> Thank you in advance for your cooperation.
>>
>> ------------------------------------------------------------------------
>
>
>
> ------------------------------------------------------------------------
>
> This message is for the designated recipient only and may contain
> privileged, proprietary, or otherwise confidential information. If you
> have received it in error, please notify the sender immediately and
> delete the original. Any other use of the e-mail by you is prohibited.
> Thank you in advance for your cooperation.
>
> ------------------------------------------------------------------------
Re: OfflineScanner
Posted by Ara Ebrahimi <ar...@argyledata.com>.
OfflineScanner is package protected. So I'll need to hack it. If it proves to be faster at least 20% then it's worth having it in the public Ali, perhaps even let user use it by a asking specific file to be scanned rather than directing scan by carefully defining the range to touch the intended file.
Ara.
On Feb 19, 2015, at 8:15 AM, Keith Turner <ke...@deenlo.com>> wrote:
On Thu, Feb 19, 2015 at 12:57 AM, Ara Ebrahimi <ar...@argyledata.com>> wrote:
Hi,
I'm trying to optimize a connector we've written for Presto. In some cases we need to perform full table scans. This happens across all the nodes but each node is assigned to process only a sharded subset of data. Each shard is hosted by only 1 RFile. I'm looking at the AbstractInputFormat and OfflineIterator and it seems like the code is not that hard to use for this case. Is there any drawback? It seems like if the table is offline then OfflineIterator is used which apparently reads the RFiles directly and doesn't involve any RPC and I think should be significantly faster. Is it so? Is there any drawback to using this while the table is not offline but no other app is messing with the table?
The code will throw an exception if the table is not offline (intent is to ensure the files are stable and not garbage collected). As others have stated you can clone.
Currently offline scanning is only supported in the public API w/ Map Reduce. Curious, would you be interested in seeing this in the client public API?
Thanks,
Ara.
________________________________
This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
________________________________
________________________________
This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
________________________________
________________________________
This message is for the designated recipient only and may contain privileged, proprietary, or otherwise confidential information. If you have received it in error, please notify the sender immediately and delete the original. Any other use of the e-mail by you is prohibited. Thank you in advance for your cooperation.
________________________________
Re: OfflineScanner
Posted by Keith Turner <ke...@deenlo.com>.
On Thu, Feb 19, 2015 at 12:57 AM, Ara Ebrahimi <ar...@argyledata.com>
wrote:
> Hi,
>
> I’m trying to optimize a connector we’ve written for Presto. In some cases
> we need to perform full table scans. This happens across all the nodes but
> each node is assigned to process only a sharded subset of data. Each shard
> is hosted by only 1 RFile. I’m looking at the AbstractInputFormat and
> OfflineIterator and it seems like the code is not that hard to use for this
> case. Is there any drawback? It seems like if the table is offline then
> OfflineIterator is used which apparently reads the RFiles directly and
> doesn’t involve any RPC and I think should be significantly faster. Is it
> so? Is there any drawback to using this while the table is not offline but
> no other app is messing with the table?
>
The code will throw an exception if the table is not offline (intent is to
ensure the files are stable and not garbage collected). As others have
stated you can clone.
Currently offline scanning is only supported in the public API w/ Map
Reduce. Curious, would you be interested in seeing this in the client
public API?
> Thanks,
> Ara.
>
>
>
> ________________________________
>
> This message is for the designated recipient only and may contain
> privileged, proprietary, or otherwise confidential information. If you have
> received it in error, please notify the sender immediately and delete the
> original. Any other use of the e-mail by you is prohibited. Thank you in
> advance for your cooperation.
>
> ________________________________
>