We have a set of our own programs to verify the performance of upgrades. Why isn’t that enough?
Many installations have a set of programs that they run on each new machine, but few installations take the time to update these programs. So, they might not reflect the actual programs in use today. And they might not invoke the instruction sets that provided most of the improved speed of the new machine. BoxScore II resolves that problem by identifying any and all stable job steps or transactions and then quantifies the changes of those specific steps and transactions.
Homegrown tools can’t tell the difference between latent demand and changes in CPU usage by job step or transaction. If you can identify the latter, you’re in a position to renegotiate your IBM contract to reflect actual value that the new CPU provides, potentially saving tens or hundreds of thousands of dollars per year. Also, homegrown tools are hard to maintain and hard for newer staff to use.