Deciding when to dump trace data is a significant decision because the dump affects system performance.
The Dump Trace (DMPTRC) command puts information from an internal trace table into a database file. It is not good to dump trace data during peak activity on a loaded system or within a high priority (interactive) job. You can delay a trace dump, but you want to dump the data before you forget that it exists. If the trace table becomes cleared for any reason, you lose the trace data. However, delaying the dump slightly and then using the DMPTRC command to dump the trace in a batch job can preserve performance for the users.
To dump trace data, issue the following command:
DMPTRC MBR (member-name) LIB (library-name)
You must specify a member name and a library name in which to store the data. You can collect sample-based data with Collection Services at the same time that you collect trace information. When you collect sample data and trace data together like this, you should place their data into consistently named members. In other words, the names that you provide in the CRTPFRDTA TOMBR and TOLIB parameters should be the same as the names that you provide in the DMPTRC MBR and LIB parameters.