[pgpool-general: 2590] Pgpool in a HA environment

[Anupama Ramaswamy]

My current setup involves a Tomcat Web Application on a Postgres backend. I need to have a load balanced / highly available application.  So I have setup postgres with streaming replication and 2 Tomcat server load balanced in a cluster. I have pgpool to load balance the postgres read-only queries.

I do not want pgpool to be a single point of failure. So I need to setup 2 pgpool instances. Here is my question.
Of the 2 configurations below, which is recommended and why ?

1) Application server = Tomcat + pgpool ---> Each tomcat connects to the local instance of pgpool which load balances against the 2 DB servers. So when one of the app servers is down, the other will continue to operate.

2) Application Server 1 = Tomcat (active) + active PgPool. Web app connects to local instance of PGpool (through Virtual IP)
Application Server 2 = Tomcat (active) + passive PgPool. The web app connects to remote instance of pgpool  (through Virtual IP). The local instance is configured as "Other PgPool" with a watchdog. This takes over only if the first instance goes down.

Not able to figure out if the first approach has any pitfalls - 2 disparate pgpools supporting 2 front-ends will cause any inconsistencies. Any help/ suggestions welcome.

Regards,
Anupama.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20140219/3382e4c2/attachment-0001.html>