We have a standard general set of monitors for all Windows servers, one of which monitors system CPU. If the CPU usage is over 90% for a period, the top 10 CPU bound process are returned in an email alert.
For some specialised servers, e.g. Exchange we have added one or more of the Exchange Application Templates.
I notice that many of the checks in these templates are checking the CPU use for individual services or processes, alerting if and individual process uses more than 90% CPU. But we will already have had a system alert that will also trigger and name the offending process(es).
It seems to me that these individual service/process CPU checks are a waste of time, or am I missing something?
Could it be that the individual checks are measuring usage against a single core, which might not trigger a system alert? But I wouldn't have thought that Exchange processes would be limited to run on particular cores, so high usage on a single core is probably not an issue.
I'm thinking of disabling the individual service CPU checks, but though I'd ask this vast knowledge base if I'm missing something.