headgeek
Contributor
Contributor

Actually these two servers had been running for about 3 months both with predictable workloads. One was running 3 minutes peaks of 100% followed by 3 minutes of idle. The other was running 6 minute peaks followed by 6 minutes of idle. So when I changed one to idle and the other to 100% all the time, by your logic it should have sent the overall health into a tailspin, however it did not. In fact It looked completely heathy even though there were clearly significant anomolies.

To further explore this logic, I simply switched the workloads. So the idle server went to 100% and the the 100% server went to idle. This did reduce the health but it still showed one of the servers as green (the one that had been red). Again this should have been considered a problem because it was not what was predicted.

Capture.PNG

So I am still left wondering what it is doing an why? I really need to be able to validate the results before we put this in production and so far it has been very hard to set up predictable tests. I haven't been able to find any documented test lab studies for the product either.