Trace logging

A casual user of the trace can use trace logging very simply by accepting all defaults. The system administrator has all the options that are necessary for complete automation and control.

Trace logging provides the following flexible reporting capabilities:

• All data for an active trace can be viewed online in the trace displays (LTRAC, STRAC and DTRAC). When you select STRAC to display current trace data, STRAC displays either current or historical data, depending on which data is most complete. If the desired data is no longer available in the online buffers, the HISTORY expand button on the trace displays (LTRAC, STRAC, DTRAC) can be used to retrieve the data from the current log data set, where it can be viewed and analyzed with all the online trace displays.

  The data that is provided in the LTRAC and STRAC displays is also available in windows mode views TRLTRAC and TRSTRAC.

• Historical trace data can be viewed online at any time. All trace log data sets are tracked in an online trace directory. The History Traces application displays these data sets and allows online recall with a simple line command, whether the trace was written the same day, last night, a year ago, or on another system. This application provides these benefits:
  • Data from completed traces is saved in the trace logs so that no data is lost during system interruptions.
  • The older data from an active trace with multiple log data sets can be accessed through this application.
  • When the current log data set for an active trace is accessed through the online trace directory, a CURRENT expand button on the trace displays provides automatic switching to displays of the online buffer data.
  • Online administration of the trace logs simplifies the tracking of which traces are valuable enough to keep for later reporting and which should be deleted after review. An additional option allows the submission of a batch job to print any trace.
  • Traces can be kept as thread profiles for later comparisons after application or system changes are made that might affect performance. A trace of an application on the development system can be compared to the same trace on production to easily spot access path changes or performance degradation that is not shown by EXPLAIN.
• Short-term workload history is always available. The workload summary displays TSUMx and TSTAT can be used online against the data from any of the trace logs to analyze workload trends or thread exceptions at a particular time or for a specific plan or user. The accounting data is always available to back up these summaries. For a detail trace the detail events are there too, already correlated for ease of use.
• Traces can be printed in batch for offline review. All trace displays and data levels shown online can also be reproduced as batch reports for hardcopy review. You can choose between a simple application summary report with one line per accounting record; all the data from that accounting record; workload summaries by authorization ID, plan, or time; or the complete detail trace of chronological events with event pop-up displays.
• All summary trace reports can be printed directly from SMF. All the displays available for a summary trace are also available as batch reports from SMF accounting records. If there is no active trace in , it is still possible to produce the equivalent reports. These reports can be produced from unloaded SMF data sets, from the current SYS1.MANx data sets, or from a GTF trace.
• Very large applications can be traced. The logging facility also enables tracing of long-running applications without loss of data because of online buffer limitations.