You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Gary Yao (Jira)" <ji...@apache.org> on 2020/02/25 09:11:00 UTC
[jira] [Commented] (FLINK-16017) Improve attachJobGraph Performance
[ https://issues.apache.org/jira/browse/FLINK-16017?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17044274#comment-17044274 ]
Gary Yao commented on FLINK-16017:
----------------------------------
{{AdaptedRestartPipelinedRegionStrategyNG}} should only be required in legacy scheduling which will be removed soon. If the removal is not possible, we can maybe avoid the creation of the {{AdaptedRestartPipelinedRegionStrategyNG}} instance, or avoid the invocation of {{notifyNewVertices}}.
cc: [~zhuzh]
> Improve attachJobGraph Performance
> ----------------------------------
>
> Key: FLINK-16017
> URL: https://issues.apache.org/jira/browse/FLINK-16017
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Coordination
> Affects Versions: 1.10.0
> Reporter: Jiayi Liao
> Priority: Major
>
> Currently {{RegionPartitionReleaseStrategy}} and {{AdaptedRestartPipelinedRegionStrategyNG}} both need to compute distinct pipelined regions, which affects the performance due to the duplicate calculating.
> The best idea that comes to my mind so far is calculating distinct pipelined regions in {{DefaultExecutionTopology}} so that we can remove the duplicate calculating.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)