[Pgpool-general] Forcing query to never hit the master

[Mark Plutowski]

To decouple long-running selects from the master.
For example, reports and data-mining queries. This avoids slowing down the
master due to someone running a report, and also avoids putting unnecessary
indexes on the master.  We direct these long-running queries to a read-only
replica that can be loaded up with indexes customized to optimize those
types of queries, without requiring writes to the master to be slowed by
those indexes. Currently we do this in the application, and I wonder if
pgpool could handle this instead.



On Tue, Jul 14, 2009 at 2:03 AM, Tatsuo Ishii <ishii at sraoss.co.jp> wrote:

> > Sorry if this has been answered in the docs, but it was not immediately
> > apparent from reading the docs.
> >
> > I'm considering pgpool to load-balance selects across one or more
> read-only
> > replicas.  I understand how to get a query to only go to the master.  Is
> > there a way to configure pgpool so that certain queries never hit the
> > master?
>
> At this point, no. However I'm curious why you need the
> functionality. And how do you specify "certain query"?
> --
> Tatsuo Ishii
> SRA OSS, Inc. Japan
>



-- 
- Mark Plutowski
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://pgfoundry.org/pipermail/pgpool-general/attachments/20090722/b8a6b782/attachment.html>