My understanding of LR is that you can drive the app from the web side and be able to drill down on the database side to find problems. I remember being shown/told that you could find the cpu spikes and drill down to the statements.
I just left a QAload pitch and they said that for them to find a bad sql statement it would require another load test between the app server and the database server. They implied that this would be the same for LR also.
Could somebody help me figure out the truth.
Drill down seems to be the point of confusion. CompuWare is correct from a tool perspective only. However, you do not necessarily need a separate test between the app server and db server to force or demonstrate db-server problems.
The drill down exercise is manual. If your performance test results reports show you that the simulated users are not performing as expected, you can manually look at db logs (as one point of investigation and not necessarily ruling out other servers) and narrow down to the queries that were running when the vusers began "complaining". The bottom line is that analyses of results and logs is necessary in order to find the fire behind the smoke.
You need to get back to your Compuware rep that talked to you and ask them about coupling QALoad with Vantage. I know for a fact, because I have done it, that if you have Vantage running on the database server along with QALoad doing a load test, you can drill down on the spikes and go directly to the SQL statement that was running on the database at the time of the spike. It is a single load test with the tools synchronized so that all the required data is in one place.