I have no idea how efficient data bases are wrt set covering, but if you want to try within Perl, there is a CPAN module using the 'greedy algorithm' (which does a pretty good approximation of the optimal set cover).