Replication. That's all. Basically, it lets you replicate data from one set of tables / sequences in one database / schema to another database / schema.
Note that slony works by using a combination of userland triggers which are used to "audit" all changes to tables and sequences in the source data set, placing these changes into intermediate tables. Then, external daemons run that talk to each other on both machines and exchange the data.
Since it is asynchronous replication, a failure of the provider could result in lost data between it and the subscriber(s).
So, we use it to make sure we have a hot, nearly 100% standby unit in case the primary server fails.
Also, one of the weirdly useful features is that you can upgrade your databases live by putting an 8.0 server behind a 7.4 server, initiating replication, and waiting for the subscriber to catch up and then switching them over at the lowest demand hour, like 2 in the morning. This also lets you preview the same changes and such on both machines, so you can see if some part of postgresql got noticeably different between versions and will affect your business. All you have to do is watch the subscriber with the new version of pgsql for problems.
