[pgpool-general: 1715] Re: Planning a cluster

[web at mr-paradox.net]

On Fri, May 10, 2013 at 08:24:07PM +0100, Garry Saddington wrote:
- Hello, I'm new here so please forgive if I ask stupid questions.
- I am planning a new db/application cluster and would like some advice. Pgpool sounds like the best tool for us but I'm still not sure.
- What we want to do is have multiple application servers in a redundant configuration and multiple synchronised Postgres servers, such that if any application or database servers go down, the rest just keep working without the need to failover. Of course we would then need to be able to fix and replace the broken servers easily.
- I initially thought of having Pgpool on each of the app servers but after doing some reading again I'm not too sure.
- Could anyone suggest the best route I can take.
- Thanks
- Garry

You're options depend somewhat on your application stack. 

However, for PgPool what you would normally do would be to have a server (or cluster for HA)
dedicated to PgPool. You would not put it on the App or DB server. it's fairly CPU
intensive. You would not load balance over, or have more than 1 active PgPool.