Your task at hand – you need to stresstest your SQL system and got yourself a .trc trace file with all the accurate transactions that defines the test. Now maybe you think you need to devide the code into bits and pieces per connection and ask a couple of guys to run each bit at a certain point in time, on different computers, or adding some WAITFOR commands and have the whole office at your own disposal, running from computer to computer to start the tests so you can have simontaneously connections with a ”spontanous” load. The problem is to re-do it in the same way on repeating test events. There you have one scenario when you rather use something like the upcoming Denalis Distributed Replay (DReplay) Utility. DReplay have a couple of different sequencing modes to orchestrate the whole test all by yourself in advance and then just run it. It has a couple of XML configuration files that you can use to specify configurations before hand like at what speed the code should run, what logging should take place, if system session activities should occur during the replay and so forth.
It introduces some components that make up the DReplay environment:
– The distributed adminstration tool
– The distributed replay controller
– One or more distributed replay clients * The target server you want to test I’m really going to look deeply into this utility when it comes to production status of the Denali code and urge you to do the same if you are the guy responsible for the stress testing of SQL Server. Until then you and I can study the preview specs and details so far at this msdn location.