Do you mean you usually disgard much of the data selected via Perl/DBI, because you grabbed too much, and need to filter it afterwards? If so I think you can do plenty more (including subqueries), using normal SQL already.

I assume PL/Perl is just 'another way to write SQL procedures'? Assuming its just as fast, it sounds interesting, but I'd just use it to write functions to be called in normal queries, if anything.

C.