Higehst CPU usage in past 15 Minutes
Higehst CPU usage in past 15 Minutes
Is there a test that can be ran which shows the Peak CPU usage in a set time period? Instead of creating a CPU usage test and running every minute-30 seconds, I would like to run every 5 minutes and pull the peak CPU usage from the machine over the past 5 minutes. Is this possible?
RRing,
The real question is, "What information are you trying to capture?" CPU spikes, depending upon your active applications and services, can happen frequently but should be short lived. For instance, your antivirus product initiates a background scan or your SQL service responds to a query - You will see a burst of CPU activity - potentially to 100%, but is that what you want to capture? Is that kind of spike significant?
So I'm curious.
Is this computer/server supposed to be sitting idle most of the time?
The real question is, "What information are you trying to capture?" CPU spikes, depending upon your active applications and services, can happen frequently but should be short lived. For instance, your antivirus product initiates a background scan or your SQL service responds to a query - You will see a burst of CPU activity - potentially to 100%, but is that what you want to capture? Is that kind of spike significant?
So I'm curious.
Is this computer/server supposed to be sitting idle most of the time?Required Counter
The systems are industrial based telecommunication servers which have burst of activity due to caller volume at different times. We are trying to capture Peak CPU(a More accurate reading), without setting the time down which could cause a system resource issue. The applications we montior, besides the OS, are all proprietary. Any idea's that would not cause significant load would be greatly appreciated.
Sorry, but I'm not familiar with your type of environment or what the goal is here. Still I'm willing to give it a try 
What OS?
I appreciate your desire to minimize the resource footprint, but generally speaking CPU monitoring is not terribly resource hungry in any environment. And as long as HM is running on a different server it's nothing more than a remote query - even the logging (writes to disk) is done on the HM server and not the target of the test.
And if you set the alarm threshold low enough you could, for instance, turn down the frequency of this test (and others) in order to service the caller volume, if that is the concern. Or maybe you initiate a processs to distribute the load to another server in the cluster.
Or maybe you want to go the other direction with it. If so is there a potential dependency you could take advantage of in order to 'ramp up' a CPU test?
For instance, query the application (or a log the application uses) for the number of active calls: While active calls > {threshold} initiate (or increase the frequency of) CPU test.
What OS?
I appreciate your desire to minimize the resource footprint, but generally speaking CPU monitoring is not terribly resource hungry in any environment. And as long as HM is running on a different server it's nothing more than a remote query - even the logging (writes to disk) is done on the HM server and not the target of the test.
And if you set the alarm threshold low enough you could, for instance, turn down the frequency of this test (and others) in order to service the caller volume, if that is the concern. Or maybe you initiate a processs to distribute the load to another server in the cluster.
Or maybe you want to go the other direction with it. If so is there a potential dependency you could take advantage of in order to 'ramp up' a CPU test?
For instance, query the application (or a log the application uses) for the number of active calls: While active calls > {threshold} initiate (or increase the frequency of) CPU test.