[
https://issues.apache.org/jira/browse/HBASE-1861?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12896082#action_12896082
]
Lars Francke commented on HBASE-1861:
-------------------------------------
Okay after having talked with Lars George and looking over the incremental load
stuff from Todd I've got even more questions.
It seems as if - and we should really document this somewhere - there are now
two distinct ways to bulk load stuff into HBase:
loadtable.rb creates regions manually and just creates the metadata to be
picked up by the metascanner. This seems like it is not very resource intensive
(after the HFiles have been generated).
And then there's the new completebulkload tool which shifts some of the load to
HBase itself by (and please correct me if I understood this wrong) possibly
splitting a lot of the existing regions and basically depending on HBase to put
HFiles in appropriate places. This is a great solution for incremental loads as
regions already exist. But is this a good solution performance/load wise for an
empty table? My knowledge of HBase in this regard is still limited but I would
have thought that the constant splitting would be pretty bad especially when
starting with an empty table with no regions.
I'd love your input on how to solve this: multi column families only for empty
tables supported by loadtable.rb or only for the incremental bulk load tool or
for both?
This also includes the question if we should keep loadtable.rb if it is a
better fit for "cold imports".
> Multi-Family support for bulk upload tools (HFileOutputFormat / loadtable.rb)
> -----------------------------------------------------------------------------
>
> Key: HBASE-1861
> URL: https://issues.apache.org/jira/browse/HBASE-1861
> Project: HBase
> Issue Type: Improvement
> Components: mapreduce
> Affects Versions: 0.20.0
> Reporter: Jonathan Gray
> Fix For: 0.90.0
>
>
> Add multi-family support to bulk upload tools from HBASE-48.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
