Tuesday, August 16, 2011

Pg-pool-II Introduction / Review of replication concepts

Now I'll be exploring yet another database replication solution oriented to Postgresql.

Pg-pool-II Introduction


  • Not only does replication but also connection pooling, load  balancing and query parallelization.

  • Supports multimaster configurations.http://www.mail-archive.com/pgpool-general@pgfoundry.org/msg02512.html

  • Its replication method is synchronous (commits are granted once all nodes have replicated the transaction) .

  • It works along with Postgresql streaming replication.


  • All the postgresql instances must be of the same major version.

  • Not advisable to use different OS and / or hardware architectures, especially if you intent to use online recovery.

  • Doesn't work with Windows

Review of replication concepts

One issue that RAF pointed me out (yet again, as he usually does)  is that synchronous replication in the proposed terms might impose a considerable extra load over transactions.If we have several nodes that are replicating data each one of them with a particular network delay, a transaction could take much time to commit. There's also a risk of our transactions failing constantly due to this overhead if our cluster architecture is not  properly planned.

No comments:

Post a Comment