[
https://issues.apache.org/jira/browse/HBASE-7841?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13592658#comment-13592658
]
Jesse Yates commented on HBASE-7841:
------------------------------------
TL;DR I'm worried we are making it more complicated than necessary and don't
gain any notable perf gains.
I know there is a comment in the code that indicates we should do this, but my
main concern here is _is it necessary_? What does the perf look like for a lot
of regions? Is that acceptable? Should be parallelize the tasks (table info
copy, region info copy) or parallelize on a per-region basis alone? How much
does it really impact if we still have to go serially through the namenode?
I would love to see some comparative numbers justifying the need before this
goes in.
> Parallelize offline snapshot in DisabledTableSnapshotHandler
> ------------------------------------------------------------
>
> Key: HBASE-7841
> URL: https://issues.apache.org/jira/browse/HBASE-7841
> Project: HBase
> Issue Type: Sub-task
> Reporter: Ted Yu
> Assignee: Ted Yu
> Fix For: 0.98.0
>
> Attachments: 7841.txt, 7841-v2.txt
>
>
> In DisabledTableSnapshotHandler, there is TODO:
> {code}
> // TODO consider parallelizing these operations since they are independent.
> Right now its just
> // easier to keep them serial though
> @Override
> public void snapshotRegions(List<Pair<HRegionInfo, ServerName>>
> regionsAndLocations) throws IOException,
> {code}
--
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
