Data sets involved in exception profiling

Exception profiling requires several data set specifications.

The following data sets are essential for exception profiling. You must specify these data sets in the exception profiling dialog.

• The input data set contains a sample of DB2 instrumentation data. The data is used to estimate your workload and decide which warning and problem thresholds to assign to the DB2 fields that are to be calculated.

  The more representative the data, the more accurate the calculation of the threshold values. The data should contain the type of data that you usually monitor. The input data should also contain a sufficient number of records to allow the profiling to be performed with reasonable confidence. The data should also cover an appropriate span of time.

  You can specify any combination of DPMOUT, SMF, or GTF data sets that contain DB2 instrumentation data.

• The input threshold data set is an Exception Threshold data set with some DB2 fields marked by asterisks (*). The asterisks serve as markers to indicate the DB2 fields for which exception profiling should calculate threshold values. Exception profiling scans the input threshold data set for names of DB2 fields that have asterisks (*) assigned instead of threshold values. For these fields new threshold values are calculated based on sample data from the input data set. Fields that already have threshold values assigned are not considered and remain unchanged.

  In other words, if you want exception profiling to calculate or recalculate certain threshold values, enter asterisks as threshold values for the corresponding fields in the input threshold data set.

• The output threshold data set is the new Exception Threshold data set. It is basically a copy of the input threshold data set (previously existing threshold values are retained) , but previously marked fields obtained new threshold values.

The output report data set is optional. If you request a profile report in the exception profiling dialog, this data set will contain a report with details of the distribution and expected number of exceptions for each DB2 field listed in the new Exception Threshold data set.