Re: Has anybody ever tried running Spark Streaming on 500 text streams?

[Tathagata Das <td...@databricks.com>]

@Brandon, the file streams do not use receivers, so the bottleneck is not
about executors per se. But there could be couple of bottlenecks
1. Every batch interval, the 500 dstreams are going to get directory
listing from 500 directories, SEQUENTIALLY. So preparing the batch's RDDs
and jobs can take a time. So your batch interval cant be small, may have to
be 10s of seconds. Which is probably fine for your application, otherwise
you would not be using files in the first place.
2. Processing new files from 500 directories may take significant
computation power. Just make sure you get large enough cluster.

On Fri, Jul 31, 2015 at 2:40 PM, Ashwin Giridharan <as...@gmail.com>
wrote:

> Thanks a lot @Das @Cody. I moved from receiver based to direct stream and
> I can get the topics from the offset!!
>
> On Fri, Jul 31, 2015 at 4:41 PM, Brandon White <bw...@gmail.com>
> wrote:
>
>> Tathagata,
>>
>> Could the bottleneck possibility be the number of executor nodes in our
>> cluster? Since we are creating 500 Dstreams based off 500 textfile
>> directories, do we need at least 500 executors / nodes to be receivers for
>> each one of the streams?
>>
>> On Tue, Jul 28, 2015 at 6:09 PM, Tathagata Das <td...@databricks.com>
>> wrote:
>>
>>> @Ashwin: You could append the topic in the data.
>>>
>>> val kafkaStreams = topics.map { topic =>
>>>     KafkaUtils.createDirectStream(....topic...).map { x => (x, topic) }
>>> }
>>> val unionedStream = context.union(kafkaStreams)
>>>
>>>
>>> @Brandon:
>>> I dont recommend it, but you could do something crazy like use the
>>> foreachRDD to farm out the jobs to a threadpool, but the final foreachRDD
>>> waits for all the jobs to complete.
>>>
>>> manyDStreams.foreach { dstream =>
>>>    dstream1.foreachRDD { rdd =>
>>>     // Add runnable that runs the job on RDD to threadpool
>>>     // This does not wait for the job to finish
>>>   }
>>> }
>>>
>>> anyOfTheManyDStreams.foreachRDD { _ =>
>>>     // wait for all the current batch's jobs in the threadpool to
>>> complete.
>>>
>>> }
>>>
>>>
>>> This would run all the Spark jobs in the batch in parallel in thread
>>> pool, but it would also make sure all the jobs finish before the batch is
>>> marked as completed.
>>>
>>> On Tue, Jul 28, 2015 at 4:05 PM, Brandon White <bw...@gmail.com>
>>> wrote:
>>>
>>>> Thank you Tathagata. My main use case for the 500 streams is to append
>>>> new elements into their corresponding Spark SQL tables. Every stream is
>>>> mapped to a table so I'd like to use the streams to appended the new rdds
>>>> to the table. If I union all the streams, appending new elements becomes a
>>>> nightmare. So there is no other way to parallelize something like the
>>>> following? Will this still run sequence or timeout?
>>>>
>>>> //500 streams
>>>> streams.foreach { stream =>
>>>>   stream.foreachRDD { rdd =>
>>>>     val df = sqlContext.jsonRDD(rdd)
>>>>     df.saveAsTable(streamTuple._1, SaveMode.Append)
>>>>
>>>>   }
>>>> }
>>>>
>>>> On Tue, Jul 28, 2015 at 3:42 PM, Tathagata Das <td...@databricks.com>
>>>> wrote:
>>>>
>>>>> I dont think any one has really run 500 text streams.
>>>>> And parSequences do nothing out there, you are only parallelizing the
>>>>> setup code which does not really compute anything. Also it setsup 500
>>>>> foreachRDD operations that will get executed in each batch sequentially, so
>>>>> does not make sense. The write way to parallelize this is union all the
>>>>> streams.
>>>>>
>>>>> val streams = streamPaths.map { path =>
>>>>>   ssc.textFileStream(path)
>>>>> }
>>>>> val unionedStream = streamingContext.union(streams)
>>>>> unionedStream.foreachRDD { rdd =>
>>>>>   // do something
>>>>> }
>>>>>
>>>>> Then there is only one foreachRDD executed in every batch that will
>>>>> process in parallel all the new files in each batch interval.
>>>>> TD
>>>>>
>>>>>
>>>>> On Tue, Jul 28, 2015 at 3:06 PM, Brandon White <
>>>>> bwwinthehouse@gmail.com> wrote:
>>>>>
>>>>>> val ssc = new StreamingContext(sc, Minutes(10))
>>>>>>
>>>>>> //500 textFile streams watching S3 directories
>>>>>> val streams = streamPaths.par.map { path =>
>>>>>>   ssc.textFileStream(path)
>>>>>> }
>>>>>>
>>>>>> streams.par.foreach { stream =>
>>>>>>   stream.foreachRDD { rdd =>
>>>>>>     //do something
>>>>>>   }
>>>>>> }
>>>>>>
>>>>>> ssc.start()
>>>>>>
>>>>>> Would something like this scale? What would be the limiting factor to
>>>>>> performance? What is the best way to parallelize this? Any other ideas on
>>>>>> design?
>>>>>>
>>>>>
>>>>>
>>>>
>>>
>>
>
>
> --
> Thanks & Regards,
> Ashwin Giridharan
>