[ https://issues.apache.org/jira/browse/BEAM-5987?focusedWorklogId=185871&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-185871 ]
ASF GitHub Bot logged work on BEAM-5987: ---------------------------------------- Author: ASF GitHub Bot Created on: 16/Jan/19 16:45 Start Date: 16/Jan/19 16:45 Worklog Time Spent: 10m Work Description: kennknowles commented on pull request #7091: [BEAM-5987] Spark: Share cached side inputs between tasks. URL: https://github.com/apache/beam/pull/7091#discussion_r248360258 ########## File path: runners/spark/src/main/java/org/apache/beam/runners/spark/util/CachedSideInputReader.java ########## @@ -86,9 +55,27 @@ private CachedSideInputReader(SideInputReader delegate) { @Override public <T> T get(PCollectionView<T> view, BoundedWindow window) { Review comment: Ah, the meaning of that comment is that `null` is a value. You can have a `PCollection<@Nullable Foo>` that contains just one copy of `null` and use `View.asSingleton()` and the side input returns the `null`. In other words, `get` _must_ return a value of type `T`. But the type `T` may itself be `@Nullable Something`. The annotation on `SideInputReader` should be removed. It is is incorrect if we use a static analysis that understands this. Findbugs does not understand this but we should aspire for our annotations to be correct so the documentation is clear. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org Issue Time Tracking ------------------- Worklog Id: (was: 185871) Time Spent: 6h 40m (was: 6.5h) > Spark SideInputReader performance > --------------------------------- > > Key: BEAM-5987 > URL: https://issues.apache.org/jira/browse/BEAM-5987 > Project: Beam > Issue Type: Bug > Components: runner-spark > Affects Versions: 2.8.0 > Reporter: David Moravek > Assignee: David Moravek > Priority: Major > Fix For: 2.9.0 > > Attachments: Screen Shot 2018-11-06 at 13.05.36.png > > Time Spent: 6h 40m > Remaining Estimate: 0h > > We did some profiling of a spark job and 90% of the application time was > spent on side input deserialization. > For spark, an easy fix is to cache materialized side inputs per bundle. This > improved running time of the profiled job from 3 hours to 30 minutes. -- This message was sent by Atlassian JIRA (v7.6.3#76005)