Viewing anomalies

When Anomaly Detection starts, it begins its learning phase. If for example, we have configured to use correlation ID for grouping, 100 occurrences of a CICS transaction (based on correlation id) would establish the average CPU, Elapsed and Get Page for that transaction. For example, 3 seconds average and 2 second standard deviation for elapsed time, .000005 average for CPU seconds and 100 average for get pages. If after the first 100 occurrences, if a transaction had an elapsed time of 14 seconds, this is more than 5 standard deviation away from the mean (5 being the tolerance), it would be reported as an elapsed time Anomaly.

1. From the OMEGAMON Enhanced 3270 User Interface, you can set the range of historical data you want to see (the default is two hours). Selecting Edit > Preferences and select the History tab where you can then change the range of data displayed for history.
2. From the Db2 main menu, select A to view any anomalies within the range selected.
3. Select S to show more details for the thread that triggered an anomaly.
   Using machine learning and artificial intelligence allows us to set smart thresholds based on what we have learned from experience. This reduces the number of false-positives where threads are flagged for exceeding a threshold but are not truly an error. In this way, we can concentrate performance tuning efforts to the specific threads that are causing performance problems.