> I want to add replication capabilities to my postgresql installation. Can
> anyone recommend a good way (open source or otherwise) to do so?
>
> Basically, I have a web tracking application that must quickly and
reliably
> store massive amounts of tracking data. Later, I run complex queries to
> analyze this tracking data. These complex queries tend to be very resource
> intensive, so I'd like to have them run on a separate machine altogether.
I
> need a simple and reliable method to 'copy' any data added to the tracking
> database to the analysis database. Preferably, I'd be able to configure
this
> copy process to defer to high priority queries or to be considerate of the
> time of day.
How closely does the replicated data have to track your live data. I my
installation I have an cron job that runs at 2:30am daily. It backs up the
database (using pg_dump) and loads it on another machine. I can then run
queries the next day on the backup machine without hurting performance on
the primary. It also has the side effect of producing the backup file for
saving to tape or whatever. It also veryfies daily that the backup file is
not corrupt because the load would fail on the secondary machine.
This is not true replication but it might serve your needs. I looked at the
replication solutions for PostgreSQL a while ago and got the impression they
were not ready for primetime. In may have changed since then.