Pgpool-II 3.0 officially released

I was using it, but stopped. It mostly worked as advertised: a proxy to a true multi-master postgresql setup. If one of the masters went down, the application wouldn't even notice, and pgPool correctly handled rolling back a transaction on both masters even if a statement only failed on one.

Advantages over mysql: higher speed under concurrent access load and safer transactions. Modifications to table metadata is fully transactional under postgresql so you can send everything through the proxy without a worry.

Downsides: automatic resynchronisation on replication failure was a pain to set up and somewhat slow on a big, busy database.

pgPool has many settings and I spent a long time trying to tweaking timeouts - a long running query would sometimes timeout on one or both masters and pgpool tries to keep your data safe and consistent and so hangs and waits for you to sort it out. (This is no worse than MySQL though.)