I've done a bit of reading up on PLM and what to make sure my understanding of the alerts is correct. For transaction failure alerts if you select to use percentage it is based on a rolling hour window and depends on which peers happened to have executed the test in that period. I'm a bit confused on how it works if you set a straight "number of peers." A support ticket says this doesn't use a rolling window so how does it determine which test runs to look at? If it's not using a window is it only looking at each test execution which would mean if you're using one peer population then it would have to be set to '1' or there would never be the second peer to trigger the alert during that run? And if you had two peer populations with the same measurements per hour would it look at each execution (one from each group) each time the evaluation for the alert occurs?
Answer by Mark B. ·
The description in your post is accurate. It works on the same concept as backbone where the state is maintained for each node (in this case peer population) and each new measurement initiates an evaluation across the peer populations.