in reply to SQL vs Berkeley

The amount of data isn't the problem here. It's how you'd be using it. If you can get it into a "hash" model (one key, one row of values), then a DBM is just fine. However, if you'll ever want to "index" some other value in the tuples, you might as well get into the full-on database instead.

-- Randal L. Schwartz, Perl hacker