with Apache::DBI each httpd process automatically opens up just
as many handles as it needs. If each of your .cgi scripts only does
one DBI->Connect then only one handle is needed. Extra handles would
only impose extra overhead. The magic that Apache:::DBI
gives you is that once it has a handle open, it doesn't close it
so your next call to a .cgi running on that httpd process will
re-use that earlier handle.