spark-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Piyush Mukati <piyush.muk...@gmail.com>
Subject Re: Structured Stream equivalent of reduceByKey
Date Fri, 27 Oct 2017 05:11:21 GMT
Thanks, Michael
I have explored Aggregator
<https://spark.apache.org/docs/latest/api/scala/index.html#org.apache.spark.sql.expressions.Aggregator>
with
update mode. The problem is it will give the overall aggregated value for
the changed. while I only want the delta change in the group as the
aggregation we are doing at sink level too.

Below is the plan generated with count Aggregator.

*HashAggregate
StateStoreSave
*HashAggregate,
StateStoreRestore
*HashAggregate,
Exchange
*HashAggregate,
*Project
StreamingRelation

we are looking for some aggregation which will avoid state
store interactions.

Also anyone aware of any design doc or some example about how we can add
new operation on dataSet and corresponding physical plan.



On Thu, Oct 26, 2017 at 5:54 PM, Michael Armbrust <michael@databricks.com>
wrote:

> - dev
>
> I think you should be able to write an Aggregator
> <https://spark.apache.org/docs/latest/api/scala/index.html#org.apache.spark.sql.expressions.Aggregator>.
> You probably want to run in update mode if you are looking for it to output
> any group that has changed in the batch.
>
> On Wed, Oct 25, 2017 at 5:52 PM, Piyush Mukati <piyush.mukati@gmail.com>
> wrote:
>
>> Hi,
>> we are migrating some jobs from Dstream to Structured Stream.
>>
>> Currently to handle aggregations we call map and reducebyKey on each RDD
>> like
>> rdd.map(event => (event._1, event)).reduceByKey((a, b) => merge(a, b))
>>
>> The final output of each RDD is merged to the sink with support for
>> aggregation at the sink( Like co-processor at HBase ).
>>
>> In the new DataSet API, I am not finding any suitable API to aggregate
>> over the micro-batch.
>> Most of the aggregation API uses state-store and provide global
>> aggregations. ( with append mode it does not give the change in existing
>> buckets )
>> Problems we are suspecting are :
>>  1) state-store is tightly linked to the job definitions. while in our
>> case we want may edit the job while keeping the older calculated aggregate
>> as it is.
>>
>> The desired result can be achieved with below dataset APIs.
>> dataset.groupByKey(a=>a._1).mapGroups( (key, valueItr) =>
>> merge(valueItr))
>> while on observing the physical plan it does not call any merge before
>> sort.
>>
>>  Anyone aware of API or other workarounds to get the desired result?
>>
>
>

Mime
View raw message