Replaying a script from VUGen in "Load mode", one transaction takes aprox. 0,5 secounds. Running the same script with one single Vuser from The Controller and localhost as loadgenertor the same transaction takes 12 secounds.
How is that possible?
I understand that TruClient VUsers have a much larger memory footprint, but with these numbers there's got to be something completly wrong.
In VUGEN during development the execution is deliberately slowed. If you want a truer representation then you need to
or you can run the script manually using command line options for MDRV. See command line flags by going to the \bin directory and executing 'mdrv' manually (if not already in the path). This will provide you with the list of driver options you may leverage for a better view of actual time to execute a single user.
TruClient is a different animal altogether.
Just a couple of suggestions:
Are you running the VUGEN-test and the controller-test from the same machine? Are they placed in the same network, do they have the same routing to the target application?
Are you running with the same browser emulation settings on in the two examples? Perhaps one of the tests is using cached resources.
