Ngone51 commented on a change in pull request #33615:
URL: https://github.com/apache/spark/pull/33615#discussion_r681570507
##########
File path: docs/configuration.md
##########
@@ -3134,3 +3134,119 @@ The stage level scheduling feature allows users to
specify task and executor res
This is only available for the RDD API in Scala, Java, and Python. It is
available on YARN and Kubernetes when dynamic allocation is enabled. See the
[YARN](running-on-yarn.html#stage-level-scheduling-overview) page or
[Kubernetes](running-on-kubernetes.html#stage-level-scheduling-overview) page
for more implementation details.
See the `RDD.withResources` and `ResourceProfileBuilder` API's for using this
feature. The current implementation acquires new executors for each
`ResourceProfile` created and currently has to be an exact match. Spark does
not try to fit tasks into an executor that require a different ResourceProfile
than the executor was created with. Executors that are not in use will idle
timeout with the dynamic allocation logic. The default configuration for this
feature is to only allow one ResourceProfile per stage. If the user associates
more then 1 ResourceProfile to an RDD, Spark will throw an exception by
default. See config `spark.scheduler.resource.profileMergeConflicts` to control
that behavior. The current merge strategy Spark implements when
`spark.scheduler.resource.profileMergeConflicts` is enabled is a simple max of
each resource within the conflicting ResourceProfiles. Spark will create a new
ResourceProfile with the max of each of the resources.
+
+# Push-based shuffle overview
+
+Push-based shuffle is an improved shuffle architecture that optimizes the
reliability and performance of the shuffle step in Spark. Complementing the
existing shuffle mechanism, push-based shuffle takes a best-effort approach to
push the shuffle blocks generated by the map tasks to remote shuffle services
to be merged per shuffle partition. When the reduce tasks start running, they
fetch a combination of the merged shuffle partitions and some of the original
shuffle blocks to get their input data. As a result, push-based shuffle
converts shuffle services’ small random disk reads into large sequential reads.
The reduce tasks are also scheduled with locality preferences of the locations
of their corresponding merged shuffle partitions, which helps to significantly
improve shuffle fetch data locality.
+
+<p> <b> Currently push-based shuffle is only supported for Spark on YARN.
</b></p>
Review comment:
Shall we also recommend the appropriate scenario here? For example, the
workload with the high load of disk I/O and long running queires. AFAIK, the
push-based shuffle doesn't perform well on small/fast queries yet.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
