[
https://issues.apache.org/jira/browse/HBASE-3727?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13776417#comment-13776417
]
James Taylor commented on HBASE-3727:
-------------------------------------
We have a good use case for this, Andrew. Arun Singh, an open source
contributor to Phoenix, is writing a bulk CSV loader for Phoenix. We'd like to
get it working efficiently for the scenario where a data table has indexes. The
indexes are themselves HBase tables (essentially partial copies of the data
table with a different row key structure) with the same column family structure
as the data table. With this feature you've added, we can do the data table and
index table creation in a single map-reduce run. Otherwise, we have to do one
map-reduce run per data table + index tables which is obviously suboptimal.
Any chance we could get this into a patch on 0.94?
> MultiHFileOutputFormat
> ----------------------
>
> Key: HBASE-3727
> URL: https://issues.apache.org/jira/browse/HBASE-3727
> Project: HBase
> Issue Type: New Feature
> Reporter: Andrew Purtell
> Priority: Minor
> Attachments: MultiHFileOutputFormat.java, MultiHFileOutputFormat.java
>
>
> Like MultiTableOutputFormat, but outputting HFiles. Key is tablename as an
> IBW. Creates sub-writers (code cut and pasted from HFileOutputFormat) on
> demand that produce HFiles in per-table subdirectories of the configured
> output path. Does not currently support partitioning for existing tables /
> incremental update.
--
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
