I have pgbouncer running on the same server, and I get over 2000 calls to a php page per minute.
Each call does inserts data digests into partitioned tables.
I also have a multi-threaded daemon connected to the same database running background operations on the data coming in.
Works flawlessly.
BTW, 20 connections is not a heavy load at all.
In my application, pgbouncer has opened about 60 backends to handle the cases where I have a high oncurrency of data coming in at once, but usually the # non-idle connections is < 10.
From: pgsql-admin-owner@postgresql.org [mailto:pgsql-admin-owner@postgresql.org] On Behalf Of Kieren Scott
Sent: Monday, August 23, 2010 8:38 AM
To: pgsql-admin@postgresql.org
Subject: [ADMIN] pgBouncer for connection pooling
Hi,
I have a web-based application (drupal) which uses PHP to make connections to a back-end postgresql 8.3 server. The application and database are on separate servers, but as we can get 20+ concurrent connections on the database I've looked at pgBouncer to try and reduce the overhead of new connections on the database. I currently have pgBouncer running on the same server as my postgres database but I was wondering whether this is a recommended setup - are the performance benefits of using a connection pooler overshadowed by the overhead of running pgBouncer on the same server as the database?
How are other people using pgBouncer? E.g. on a separate server from the database etc...?
Thanks in advance.