I run several databases on a windows server all using the same cluster.
Each database is backed up separately using pg_dump - but the size of
the databases is leading me to think that the WAL backup strategy will
reduce my backup bandwidth. The flaw in this is that the WAL backup is
not confined to a single database but the entire cluster so that I have
to restore all the database in the cluster to a point in time, rather
than just a single database.
One method where I can still use the WAL backup is to put each database
into its own cluster. So I can restore each database individually.
Anyway, this leads me to the question: Has anyone tried splitting
multiple database (10+) into clusters and what are the memory and
processor overheads in comparrison to more connections in a single cluster?
Thanks.
Howard.
www.selestial.com