Sometimes, because of specific bottleneck factors, you need to focus only on specific time intervals. EPILOG includes a very efficient means of accomplishing this, using its exception analysis techniques.
It can be very tedious to search through hundreds of collection intervals looking for periods when a particular type of bottleneck was significant. To relieve this problem, the DISPLAY command accepts exception filters, which display only intervals that meet your selection criteria. For example, you can use exception filters when you want to see only intervals where database I/O waits accounted for more than 20% of the thread’s lifetime.
Exception filtering is, of course, optional; if you do not supply an exception filter in the DISPLAY command, all intervals within the specified time limits are processed and displayed.
If you specify multiple exceptions, all criteria must be met for an interval to be selected (the criteria are logically ANDed together).
This section discusses the exception keywords for the DISPLAY command and the exceptions for combined intervals. Table 11 lists the exceptions you can use. “Bottleneck Analysis Execution States” on page 197 describes these exceptions in detail.
Displaying EPILOG Data