[Pgpool-general] connection pooling

[Mark]

On Thu, Oct 8, 2009 at 10:47 PM, Josh Berkus <josh at agliodbs.com> wrote:

>
> > However, when I start one connection with psql, I see a pgpool and a
> > postgresql process for that connection.  I leave that one open and start
> > a new connection in the same way and now there are two pgpool processes
> > and two postgresql processes with the same properties.  How is that
> > reusing anything?
>
> It only reuses the connection if you drop the first connection.  pgPool
> is really more of a load balancer than a pooler.
>
> --Josh Berkus
>

oh, so the savings is that pgpool leaves the connection open to the backend
even after the client drops it so that the next client request can reuse the
abandoned connection rather than firing a new one up from scratch?  The
documents seem to imply otherwise, hence my confusion.  So I'll still need
to support the same number of connections to my postgres server as before.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://pgfoundry.org/pipermail/pgpool-general/attachments/20091008/753097d6/attachment.html>