What makes you think you need database connection pooling? Apache::DBI gives persistent connections, which solves the speed problem. It's easy to use, and whatever caveats you've heard are probably exaggerated. Unless some of your mod_perl processes are handling requests that don't involve a database at all, they should hold onto their connections.

Doing actual sharing of database handles is hard, since it involves passing sockets and file handles around between processes safely. That's why it hasn't been done.