> IMHO the size of the DB is less relevant than the query workload. For
> example, if you're storying 100GB of data but only doing a single
> index scan on it every 10 seconds, any modern machine with enough HD
> space should be fine.
I agree that the workload is likely to be the main issue in most
situations. However, if your queries involve lots of counting and
aggregating, your databases contains several gigabytes of data, and you
are using common hardware, be prepared to wait anywhere from minutes to
hours, even if you are the only user.