When active, this feature flushes trace records to disk before the trace table is overwritten.
Because it runs while the system is running, it can affect the performance of the system, and is therefore very configurable in order to record the specific types of events that might be needed for different types of problems. Ideally, production systems should be able to perform adequately with at least the minimum tracing enabled.
For auxiliary traces, you have the option in Enterprise Server, to write trace output to either an A or a B dataset file. The default is A, but you can toggle between the two.
Trace output files are written to your system directory, and take the following form:
To archive the auxiliary trace file, set the archive_auxiliary_trace property to true. Once set, every time you toggle between writing to casauxta.rec and casauxtb.rec the currently active .rec file is moved to a new archive file, named casauxt.nnn. Where the nnn file name component is an incremental count of the switches that have previously occurred, starting from 001. The archive file is reset to casauxt.001 on the thousandth switch.
You can set the archive_auxiliary_trace property in the Enterprise Server Common Web Administration (ESCWA) interface on the Advanced Region Properties page.