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-13894) Avoid visitor alloc each call of ByteBufferArray get/putMultiple()
Date Fri, 12 Jun 2015 15:49:00 GMT

    [ https://issues.apache.org/jira/browse/HBASE-13894?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14583591#comment-14583591
] 

stack commented on HBASE-13894:
-------------------------------

+1 on this patch... 

In your profiler, can you see where time is being spent inside multiple?

On it being a hotspot, I wonder if it inlines (it looks like a large method). Would be interesting
to see output with below enabled

 -XX:+UnlockDiagnosticVMOptions -XX:+PrintInlining -XX:+PrintCompilation



> Avoid visitor alloc each call of ByteBufferArray get/putMultiple()
> ------------------------------------------------------------------
>
>                 Key: HBASE-13894
>                 URL: https://issues.apache.org/jira/browse/HBASE-13894
>             Project: HBase
>          Issue Type: Improvement
>          Components: regionserver
>    Affects Versions: 2.0.0, 1.0.1, 1.1.0
>            Reporter: Matteo Bertozzi
>            Assignee: Matteo Bertozzi
>            Priority: Minor
>             Fix For: 2.0.0, 1.1.1
>
>         Attachments: HBASE-13894-v0.patch
>
>
> ByteBufferArray getMultiple()/putMultiple() creates a new Visitor each call. we can move
that visitor out.
> also, ByteBufferArray.multiple() seems to be an hotspot, I tried to do some experiments
using slice and read/write lock instead of a mutex, doesn't seems to change much. any thoughts?



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message