[
https://issues.apache.org/jira/browse/HBASE-17049?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16268646#comment-16268646
]
ramkrishna.s.vasudevan commented on HBASE-17049:
------------------------------------------------
Tried with 24 threads as [~zghaobac] did and also 50 threads.
Still it is 1 node DN and RS ( as I don have multi nodes readily available).
And yes I can see improvement with this patch. One more change is that I
performed this test on tip of branch-2.
For previous results (that was based on 2.0-alpha4 refer here.
https://issues.apache.org/jira/browse/HBASE-16890?focusedCommentId=16264057&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16264057
Now with both 24 and 50 threads I can see AsyncWAL performs better with this
patch (with autoFlush= true).
50 threads
FSHLog : Min: 683317ms Max: 692145ms Avg: 686008ms
AsyncWAL : Min: 515266ms Max: 522350ms Avg: 517548ms
24 threads
FSHLog : Min: 716638ms Max: 719082ms Avg: 718069ms
ASyncWAL : Min: 552599ms Max: 554915ms Avg: 553826ms
Results seems to be much better than the previous results now. Thanks
[~Apache9]. Patch LGTM.
> Do not issue sync request when there are still entries in ringbuffer
> --------------------------------------------------------------------
>
> Key: HBASE-17049
> URL: https://issues.apache.org/jira/browse/HBASE-17049
> Project: HBase
> Issue Type: Sub-task
> Components: wal
> Reporter: Duo Zhang
> Assignee: Duo Zhang
> Priority: Critical
> Fix For: 2.0.0-beta-1
>
> Attachments: HBASE-17049.patch, delay-sync.patch
>
>
> https://issues.apache.org/jira/browse/HBASE-16890?focusedCommentId=15647590&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-15647590
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
