If my memory is correct, there're a few items that Stack pointed out which
are still outstanding for
HBASE-3996<https://issues.apache.org/jira/browse/HBASE-3996>
.

Cheers

On Tue, Jul 17, 2012 at 1:39 PM, David Koch <[email protected]> wrote:

> Hello,
>
> I came across this ticket for multiple table scans via and their use in
> Map/Reduce jobs:
>
> https://issues.apache.org/jira/browse/HBASE-3996
> https://reviews.apache.org/r/4411/diff/7/
>
> There is a patch for this now and it is mentioned in the comments that the
> functionality could be useful for doing joins as part of a map reduce.
> Could someone briefly explain how this works? I am interested in doing
> joins between 2 tables on rowkeys.
>
> If I append both tables to the newly added MultiTableInputCollection
> instance and use that in a Map/Reduce - would map(<rowkey>, <value>) only
> be called once per unique <rowkey> with <value> containing 2 value sets if
> the key was found in both tables?
>
> If there exist any practical examples for doing joins on HBase tables I'd
> appreciate a link. Also, I am using Hbase client 0.90.6-cdh3u4, is the
> patch applicable to this version of HBase at all?
>
> Thank you,
>
> /David
>

Reply via email to