Some thoughts:
* as mentioned on SO, New Relic is a good place to start. Figure out where the time is going in the slow requests, and work from there.
* the rewrite to an RDBMS may help, may not. It's highly dependent on why CouchDB was chosen in the first place - if your application depends heavily on some of the features of CouchDB (MapReduce, or async replication) it may make things worse.
* if minor parts are hard to port from Couch, skip them. Get something working with the new design, and benchmark it.
* Above everything else, MEASURE. Most of the things you listed on StackOverflow *could* improve performance, but *could* decrease it.
--Matt Jones