The Java test was made with just 5 seconds of "ramp-up" time. Perl was running under the same conditions. Is that fair?

Depending on your setup, it could take the perl version longer than that to open a connection for every apache sub-process. That would slow things down considerably.

It's also possible that your apache subprocesses don't handle many requests before they are replaced by fresh ones (which also means opening a new database connection).