hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "stack (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-9416) Use the disruptor library in the rcpServer
Date Wed, 04 Sep 2013 13:45:52 GMT

    [ https://issues.apache.org/jira/browse/HBASE-9416?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13757767#comment-13757767
] 

stack commented on HBASE-9416:
------------------------------

Thats coolio.  So, instead of a set of handlers, we have the disruptor doing handoff?  Was
there an issue if the disruptor ring was not big enough and if we were not servicing it fast
enough, it could overwrite outstanding events with new ones (maybe I imagined this)?  Or does
it block before it overruns the old entries?
                
> Use the disruptor library in the rcpServer
> ------------------------------------------
>
>                 Key: HBASE-9416
>                 URL: https://issues.apache.org/jira/browse/HBASE-9416
>             Project: HBase
>          Issue Type: Improvement
>          Components: regionserver
>    Affects Versions: 0.98.0
>            Reporter: Nicolas Liochon
>            Assignee: Nicolas Liochon
>         Attachments: disrupt.patch
>
>
> Nasty & hacky patch on top of the 0.96 to get some feedback on adding this third
party.
> I ran a test doing "gets" on an empty region.
> With the current implementation, we're spending time in the LinkedBlockingQueue#put.
I was able to do 150K operations per second.
> Using the disruptor allowed me to go to 190 ops/s, i.e. a little be more than a 25% improvement.
> Likely there are other improvements in this class as well.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message