[Pgpool-hackers] High Availability for PGPool Itself

[Guillaume Lelarge]

Le 29/03/2011 05:53, Tony Hoang a écrit :
> Hi All,
> 
> I have successfully setup PGPool II to perform replication and load balancing
> and this is a great piece of middleware software.
> 
> My only concern is for high availiabity for PGPool itself, since if that
> goes down then effectively we will have to manually bring it back up and
> cause an outage.
> 
> I wanted to know if a duo setup of PGPool would work.  Say the following
> structure:
> 
> App. Server 1 -------> PGPool 1 -------> PostgreSQL 1
>               \     /           \     /
>                \   /             \   /
>                 \ /               \ /
>                  /                 /
>                 / \               / \
>                /   \             /   \
>               /     \           /     \
> App. Server 2 -------> PGPool 2 -------> PostgreSQL 2
> 
> So the application servers are configured to have two datasources and then
> can use either one depending on availability.
> 
> Both PGPool servers are setup to communicate and replication/load balancing
> the two (or more) PostgreSQL servers.
> 
> I will attempt to test this as a proof of concept but has anyone tried this
> or know whether or not there will be problems with this setup?
> 

You should have one pgpool on each app server. It works great.

Or you can have two pgpool servers and use an virtual IP address that
switch between them if one goes down.


-- 
Guillaume
 http://www.postgresql.fr
 http://dalibo.com