Although I agree the ORM layer really accelerates how quickly you can get benefit out of memcached, it is certainly incorrect that mod_perl is required to get any benefit. The initial testing has already shown modest benefit.
As another example, you have a fully parsed MARC::Record object representation for a given biblio. You can store that object away, and the next time you need to reference it (when a user clicks for the details page, or the MARC page, etc.) you can retrieve and use it without having to call:
- biblioitems for the marcxml
- MARC::Record->new_from_xml
That example may not be something we do in practice (better targets abound), but you can see conceptually where the benefit can kick in w/o mod_perl. In particular, if you have a lot of Apache front-ends, you can use a consolidated cache server and get the benefit across all (one copy of the object available to all of them).
At the end of the day, I think we are very close to seeing real production-scenario benefit from memcached, and still somewhat far from using mod_perl. In large part that is because mod_perl requires broad changes to the codebase before the first benefit is returned whereas memcached can slip in optionally now.
--Joe