[
https://issues.apache.org/jira/browse/DISPATCH-2250?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17436102#comment-17436102
]
ASF GitHub Bot commented on DISPATCH-2250:
------------------------------------------
kgiusti opened a new pull request #1407:
URL: https://github.com/apache/qpid-dispatch/pull/1407
This improves single-router throughput by about 8% for small message
traffic. The extra poll picks up any output work scheduled by the
core since the last PN_CONNECTION_WAKE event.
--
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]
> Add extra poll of pending output work on I/O thread
> ---------------------------------------------------
>
> Key: DISPATCH-2250
> URL: https://issues.apache.org/jira/browse/DISPATCH-2250
> Project: Qpid Dispatch
> Issue Type: Improvement
> Components: Router Node
> Affects Versions: 1.17.0
> Reporter: Ken Giusti
> Assignee: Ken Giusti
> Priority: Major
>
> See
> [https://github.com/kgiusti/dispatch/commit/c546f31af8a9638b6fba4ccca68edfa1adaf0498]
>
> Performance tests have shown a increase in throughput with this patch. The
> core thread will forward messages faster than the output thread can write
> them to proton. Testing has showed that when under load additional output
> work items are added to the output connection during the time it spends
> draining proactor events. This extra output poll will service those work
> items before returning control to proton to perform the I/O.
>
> There probably is a cleaner way to implement this - including a way to make
> checking for the presence of output work cheaper....
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
