While somewhat anecdotal, we looked at the previous version of the
application (based on Eclipse 3.3) on the same server environment and it
generally seemed faster -- still not good, but not as bad as yesterday's
test with the current version. [I say "somewhat anecdotal" because we
can't control what else is happening on the server, so it is possible the
additional load today was lower than yesterday.]
Based on a single, casual data point, 3.4 seems to be slower than 3.3
(though I still can't imagine it accounts for all that is going on).