[infinispan-issues] [JBoss JIRA] (ISPN-6494) Investigate bundler performance

[Johno Crawford (Jira)]

    [ https://issues.jboss.org/browse/ISPN-6494?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13772885#comment-13772885 ] 

Johno Crawford commented on ISPN-6494:
--------------------------------------

[~belaban] Thanks for clearing that up, I was a little confused why all the sample Infinispan JGroups configurations use no-bundler but it's not present in any of the JGroups sample configurations.

In the Orbit [1] benchmark I didn't notice any regressions with performance after disabling message loopback and comparing no-bundler / transfer-queue so I think we'll go ahead and enable that, the pref test didn't show any improvements either but I suspect that's because there is only a single node and the parallelism is low.

My plan now is to export the JGroups mbeans so we can follow those and see what we might be able to tune further to improve performance.

Thanks again!
[1] Orbit is a virtual actor framework that combines Infinispan / JGroups messaging as I described above.

> Investigate bundler performance
> -------------------------------
>
>                 Key: ISPN-6494
>                 URL: https://issues.jboss.org/browse/ISPN-6494
>             Project: Infinispan
>          Issue Type: Task
>          Components: Core
>    Affects Versions: 9.0.0.Alpha1
>            Reporter: Dan Berindei
>            Assignee: Dan Berindei
>            Priority: Major
>             Fix For: 10.0.0.Final
>
>
> For ISPN-6027 we changed the default JGroups bundler to {{sender-sends-with-timer}}, because it was faster in some of the performance tests. However, IspnPerfTest shows {{transfer-queue-bundler}} to be consistently better, so we need to investigate the bundler choice again.



--
This message was sent by Atlassian Jira
(v7.12.1#712002)