[
https://issues.apache.org/jira/browse/HBASE-940?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12641804#action_12641804
]
Jean-Daniel Cryans commented on HBASE-940:
------------------------------------------
Unfortunately object creation may become a real problem in a big cluster with many writes if this patch is applied.
If TableReduce was not an interface, it would have been a good place to put this copy stuff since it's only useful for MR.
On the other hand, doing it in commit prevents weird situations where a user would have reused the same BatchUpdates over
and over and only the latest update would have been applied... many times.
> Make the TableOutputFormat batching-aware
> -----------------------------------------
>
> Key: HBASE-940
> URL: https://issues.apache.org/jira/browse/HBASE-940
> Project: Hadoop HBase
> Issue Type: Improvement
> Components: mapred
> Reporter: Jean-Daniel Cryans
> Assignee: Jean-Daniel Cryans
> Fix For: 0.19.0
>
> Attachments: hbase-940-v1.patch, hbase-940-v2.patch
>
>
> With HBASE-748, we can batch updates but the auto-flushing is enabled by
> default. Make it that it is disabled for MR jobs outputting in a table and
> make sure that everything is sent in the end.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
