How does BoxScore II work?
BoxScore II is based on the premise that at least some workloads are fairly consistent. These workloads tend to use the same amount of CPU per I/O or per transaction, within a certain coefficient of variance. Using various SMF records, BoxScore II analyzes hundreds or thousands of job steps or transactions to identify the stable workloads, calculate their average CPU usage before a change, and then compare that to their CPU usage after the change.
There are three types of analysis provided: average change, median change, and changed weighted by CPU time. These three can be compared to help eliminate outliers and provide more accurate results. Using this data, BoxScore II then applies analysis based on the work of performance expert Cheryl Watson to locate the cause of the issue and suggest actions to take.
Another BoxScore II feature compares the actual performance to the expected performance based on zPCR or other estimates.