Once you have identified how to recreate an application performance problem consistently, you must collect the performance profile data surrounding the problem. In a production environment, this will involve importing the relevant data from the Tivoli Enterprise⢠Monitoring Server (TEMS). In a development or testing environment, you will turn on the profiler and record the flow of execution within the application.
Note: It is important to exercise the smallest possible portion of the application in order to focus only on the part that exhibits the problem. This will simplify the analysis later on. If you are unsure of what the trigger might be, you may have to collect more data, but use more thorough filters to view the data.
Prerequisite: The data collection infrastructure must be installed on all the machines from which you are interested in collecting data. For details on installing the infrastructure, see the toolkit installation guide in the toolkit installation directory. Also, see Troubleshooting and unsupported scenarios for information about data collection limitations.