[
https://issues.apache.org/jira/browse/PHOENIX-1675?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14326906#comment-14326906
]
James Taylor commented on PHOENIX-1675:
---------------------------------------
HBase client will parallelize batched mutations. I believe if we're using a
DirectExecutorService, we'd lose this. Since all batched writes are isolated to
MutationState, I was thinking to provide either another thread pool for writes
(i.e. QueryServices would have another new getWriteExecutor() method), or
perhaps the one we use now (rather than this new one), so that we continue to
get parallelism for writes. We handle parallelism ourselves for reads, so a
DirectExecutorService makes sense for that.
> Improve Phoenix client threading
> --------------------------------
>
> Key: PHOENIX-1675
> URL: https://issues.apache.org/jira/browse/PHOENIX-1675
> Project: Phoenix
> Issue Type: Bug
> Reporter: James Taylor
> Attachments: PHOENIX-1675_wip.patch
>
>
> See PHOENIX-971 for some interesting threading discussions. Moving here for
> further discussion.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
