[infinispan-issues] [JBoss JIRA] (ISPN-6580) Hotrod performance regressions after ISPN-5342 ISPN-6545

[Gustavo Fernandes (JIRA)]

    [ https://issues.jboss.org/browse/ISPN-6580?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13204006#comment-13204006 ] 

Gustavo Fernandes edited comment on ISPN-6580 at 5/11/16 1:05 PM:
------------------------------------------------------------------

Enabling Epoll on Netty gives considerable savings in allocation rates (66% less allocations) and perf gian, about 10% on gets and puts (measured with the reproducer test).

I've added a System property {{-Dinfinispan.server.channel.epoll=true}} to enable it. 
Epoll only works on Linux, and once we can accurately detect with ones are supported (I had reports it did not work on RHEL 6.5, further investigation needed), we can think in enabling it by default.


was (Author: gustavonalle):
Enabling Epoll on Netty gives a considerable savings in allocation rates and performance. I've added a System property called {{-Dinfinispan.server.channel.epoll=true}} to enable it. 
Epoll only work on Linux, and once we can accurately detect with ones are supported (I had reports it did not work on RHEL 6.5, futher investigation needed), we can think in enabling it by default.

> Hotrod performance regressions after ISPN-5342 ISPN-6545
> --------------------------------------------------------
>
>                 Key: ISPN-6580
>                 URL: https://issues.jboss.org/browse/ISPN-6580
>             Project: Infinispan
>          Issue Type: Bug
>          Components: Remote Protocols, Server
>            Reporter: Jakub Markos
>            Assignee: William Burns
>         Attachments: jfr_recordings.zip, pom.xml, Reproducer.java
>
>
> There were 2 recent regressions in hotrod performance, one between commits dd5501c5e and 628819461 and the second one between 628819461 and db0890270. I didn't look for the exact commits, so the name of the issue might not be 100% exact...
> It is easily reproducable locally with a single server instance, reproducer attached.
> The numbers on my machine:
> ||Build commit||Puts time||Gets time||
> |dd5501c5e|21|74|
> |628819461|26|102|
> |db0890270|48|224|
> The JFR recordings (attached, captured is only the part of the test with gets) for db0890270 show a lot of time is spent in HotRodDecoder#resetNow(), and also the allocation rate goes from 100MB/s for dd5501c5e to over 1GB/s for db0890270. There are no glaring differences between dd5501c5e and 628819461.



--
This message was sent by Atlassian JIRA
(v6.4.11#64026)