Whatever be the number of indexes on a user table, there will be exactly
one index table for that user table. All the index meta data will be stored
within this single index table. So for one region there will be always one
buddy region, which the balancer has to colocate.

When user table is created there will be index table created with same no#
regions and same region start/end keys. What is selected as the RK for the
index entries is doing the work.

It will be
Index region start key + Index name + indexed col value(s) + user table rk

-Anoop-

On Thu, Jan 9, 2014 at 9:29 PM, rajesh babu Chintaguntla <
chrajeshbab...@gmail.com> wrote:

> Hi Sudarshan,
>
> 1-1 association between user regions and index regions maintained by using
> custom load balancer(SecIndexLoadBalancer) which will always colocate user
> region and corresponding index region on the same region server.
> Let's suppose if we create an indexed table with split keys {a,b,c} then we
> will create corresponding index table with same split keys.If a user region
> with start key 'a' is assigned or assining to a region server RS, to assign
> the index region with start key 'a', balancer gives the same RS as
> destination and vice versa.
>
> Thanks,
> Rajeshbabu.
>
> On Thursday, January 9, 2014, Sudarshan Kadambi (BLOOMBERG/ 731 LEXIN)
> wrote:
>
> > Could you explain how the 1-1 association between user and index table
> > regions is maintained. I wasn't able to understand fully from the
> document.
> >
> > ----- Original Message -----
>  > From: Ted Yu <dev@hbase.apache.org <javascript:;>>
> > To: dev@hbase.apache.org <javascript:;>
> > At: Jan 8, 2014 3:41:40 PM
> >
> > Hi,
> > Secondary index support is a frequently requested feature.
> >
> > Please find the updated design doc here:
> >
> >
> https://issues.apache.org/jira/secure/attachment/12621909/SecondaryIndex%20Design_Updated_2.pdf
> >
> > HBASE-9203 is the umbrella JIRA.
> >
> > Implementation patch was attached to HBASE-10222
> >
> > Thanks to Rajesh who works on this feature.
> >
> > Cheers
> >
>

Reply via email to