March 15, 2011 at 11:11 pm
Dear All,
we are having sql serevr 2005 servers.The below problem is in all servers.
In our environment we will get alerts for cpu usage if it is more than 80%.Form 3 days onwards we are getting same alerts at different timings.when i checked in performance there it is showing normal usage only.Can you please suggest other procedures to find why this cpu usage is high.
Waiting for a quick response
Thanks in advance...
March 15, 2011 at 11:54 pm
Do you receive the alerts in the same time period each day?
Setup a trace in profiler and some perfmon capturing for that time frame to see where and when your CPU spikes and trace the processes responsible for it.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
This thing is addressing problems that dont exist. Its solution-ism at its worst. We are dumbing down machines that are inherently superior. - Gilfoyle
March 16, 2011 at 12:05 am
Not exactly same time.We are getting at different timings.Not for only one server.For all servers same issue.
March 16, 2011 at 5:46 am
It sounds like you're getting spikes on usage. That can be normal. A particularly nasty query used the CPU a lot and then cleared. The problem with setting a threshold and alert to a single value is that the least blip across the line can fire the alert. 80%+ spikes are ok as long as they aren't sustained. I wouldn't worry unless you had sustained usage at or above 80%.
"The credit belongs to the man who is actually in the arena, whose face is marred by dust and sweat and blood"
- Theodore Roosevelt
Author of:
SQL Server Execution Plans
SQL Server Query Performance Tuning
March 16, 2011 at 6:19 am
You can try checking your DMV's. here is a good link that should get you started..
March 16, 2011 at 12:20 pm
One problem my workplace has with alerts and notifications. Email can be slow across the network (so can pages), which means we get notified of an event 2-10 minutes after it occurred and is already cleared out.
If you can set up your alerts to log information somewhere as the event is occurring, that would assist you with tracking down your spike problem.
Viewing 6 posts - 1 through 5 (of 5 total)
You must be logged in to reply to this topic. Login to reply