Great article by the way...
I still exist in the SQL 2000 world. I run transactional replication. The way we're set up, we use the stored procedures for our application to maintain teh data at the subscriber. As you might imagine, as long as the application runs perfectly, there are no problems. But once you introduce an occasional deadlock, or an error on an insert etc, things can get out of sync.
I have been testing the use of the sp_msins, sp_msdel, sp_msupd schema that can be created when setting up replication. These will eliminate the possibility of errors/deadlocks in the application causing inconsistency.
My issue is performance.... and my test-side hardware etc is never going to give me a good comparison. Are there performance issues with the setup I'm trying? Just want to get some comfortability before I move to production. Any assistance/info is appreciated.