is there any benchmark/whitepaper which lists the benefits in quantitative terms for moving an computational intensive two tier architecture application to three tier architecture application.
I suspect the advantages would depend entirely on details of the application. The obvious point being that intensive computation can be carried out on multiple back-end machines managed by your middle tier. The tradeoff involves extra communication delays - object serialization, etc versus the time spent in actual computation.
Give us a hint - how much time does one "job" take - seconds, minutes, hours, days???
Bill