[
https://issues.apache.org/jira/browse/ACCUMULO-119?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13153008#comment-13153008
]
Keith Turner commented on ACCUMULO-119:
---------------------------------------
I ran some continuous ingest experiments on a 10 node cluster with 10 ingest
client. 1.4 w/o group commit was running at round 300,000 key-values/sec after
running overnight. 1.5 w/ group commit was running at around 530,000
key-values/sec after running over night. So group commit made a significant
improvement for continuous batch writes.
> Implement group commit for write ahead logs
> -------------------------------------------
>
> Key: ACCUMULO-119
> URL: https://issues.apache.org/jira/browse/ACCUMULO-119
> Project: Accumulo
> Issue Type: Improvement
> Components: tserver
> Reporter: Keith Turner
> Assignee: Keith Turner
> Fix For: 1.5.0
>
> Attachments: ACCUMULO-119-2.patch
>
>
> Currently access to the tablet server write ahead logs is synchronized. It
> may be better if all writes bound for the write ahead log were put on a queue
> and processed by a background thread. This would allow writes from multiple
> clients to be grouped and written to the write ahead log at once.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
