The explanation why there are some many virtual users required (around 6000 when there are only about 150 concurrent Oracle sessions active at any one time). BMF capture will create one user scenario/session for every Oracle session. So every time Oracle user logon, executes some SQL, and logoff is equivalent to on BMF user session. And as you know each BMF user session requires a VU to execute. So the number of virtual users required will be the total user sessions, therefore in the user’s case, Oracle saw 6000 users’ login and execute some SQL. Now as part of the capture process BMF retains the relative timing between user sessions and SQL, that way the replay will be the same as the capture. For the user session, BMF adds keying time so that the session doesn’t start until after the keying time. This keying time comes from the difference between the time the Oracle session started (logon) and when the capture started. This way the BMF replay has the same number of concurrent users active in Oracle as during capture.
Now we have had questions why we don’t combine the user sessions into fewer sessions and, therefore require less virtual users. Doing this is very complicated due to several factors like credentials during the session, relative timing between sessions and SQL calculation during combination, Oracle client library initialization issues. So the way it currently behaves was by design. It also allows the replay of the sessions to be done much quicker since much of the initialization can be done before the replay actually begins. The user session still logons to Oracle at the same relative timing. Just wanted to stress that the logon is not part of the initialization process.
© 2025 Quest Software Inc. ALL RIGHTS RESERVED. Terms of Use Privacy Cookie Preference Center