With all the performance implemented in 1.9.x and 2.x series... it seems that we aren't repeating too much queries, so rcache real benefits should be checked in order to decide if we want to continue supporting it.
It seems that internal (static, per request) rcache doesn't give us too much, but both memcached and eacellerator ones (shared) could have their point about continue being supported.
So I would propose to:
- Build some "simulation" test.
- Execute it, under concurrence in these situations:
1) rcache off
2) rcache, internal, on
3) rcache, memcached, on
4) rcahce, eacellerator, on
And for all the 1-4 test measure:
1) DB queries.
2) cache hits
3) Execution time
Feel free to add more ideas about this.... we should be able to have some good numbers in order to decide... ciao