I would recommend looking at the DBI interface and the DBD::AnyData or DBD::CSV drivers to see if they will fit your needs.

Neither of those drivers support indexes (indexes are generally implemented through the DB engine itself, text files not being a real DB) though, you'll need to implement indexes yourself (not a very fun proposition, and almost impossible to match the effiecency of a real DB)

I would highly recommend either:

PostgreSQL

MySQL

Both are free (as in beer), Open Source, and available Win32/*nix.

grep

grep> cd /pub grep> more beer