[
https://issues.apache.org/jira/browse/SPARK-24541?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16757985#comment-16757985
]
Jungtaek Lim commented on SPARK-24541:
--------------------------------------
Continuous processing requires "single stage" to let all tasks launching all
the time, as well as processing never stops. It works with query which only has
map-like transformations, but stateful operations normally requires
hash-partitioning so Spark should have some mechanism to route rows directly
(as same as other streaming frameworks are doing).
Some of works were done around more than half a year ago so I cannot remember
correctly, but the POC-like work is done with RPC, and this issue tracks the
effort of changing them to TCP.
> TCP based shuffle
> -----------------
>
> Key: SPARK-24541
> URL: https://issues.apache.org/jira/browse/SPARK-24541
> Project: Spark
> Issue Type: Sub-task
> Components: Structured Streaming
> Affects Versions: 2.4.0
> Reporter: Jose Torres
> Priority: Major
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
