Thoughts from the EPS Windows Server Performance Team
Troubleshooting High CPU utilization can be addressed from a couple of different standpoints. In the first instance, you might have a single process consuming CPU - essentially a runaway process. However, what happens if there is no runaway process, and yet the CPU utilization still continues to run high, sometimes at 100% for an extended period of time? You may be experiencing a processor bottleneck instead. Processor bottlenecks occur when the processor is so busy that it cannot respond to requests for a significant period of time.
The major indicators of a processor bottleneck are identifiable using Performance Monitor. There are two main counters to be aware of:
Interpretation of these two counters should be performed with the following considerations:
A combination of high processor utilization and a lengthy processor queue may be an indication that the processor is overloaded. Remember however, that a misbehaving application can cause the same symptoms. For example, if an application thread is caught in an infinite loop, that can drive processor utilization to 100% (although you would most likely catch this in Task Manager). Additionally, remember that sudden spikes may create a scenario where it appears that you have a processor bottleneck. Having a performance baseline of the server is key to understanding and detecting abnormal conditions and take corrective actions in a timely fashion. Remember that any normal workload that drives the CPU to 100% utilization on a consistent basis for an extended period of time should be investigated. Even if the queue of Ready threads is low, there is obviously some pressure on the CPU. A faster processor would provide relief. If the workload is multi-threaded, moving from uniprocessor to multiprocessor may offer additional relief. Dividing the workload among a cluster of machines may be another option. Lastly, consider this scenario: If the workload is something that can be shifted to a different time when there is less work being done on the machine, that may help to ease the pressure on the CPU - a common example would be running backups during business hours. These operations can cause severe I/O and Network pressure as well as CPU utilization - moving them to the evenings / weekends helps to reduce the load during business hours.
That brings us to the end of this post. In our next post, we'll cover the different ways in which a processor executes instructions. Until next time ...
- CC Hameed
PingBack from http://www.ditii.com/2008/01/15/overview-of-processor-bottlenecks/
nice post, looking forward to next one
As a New Year's Resolution I'm planning to blog at least once a week. Hopefully I will maintain this
What does it mean when consistently high PQL's coincide with consistently low CPU utilizations?
My instinct would be to suggest "poorly written code" as an answer, but I'd like some info to either back up that info or refute it.
THIS IS THE NIDE ARTICLE
Hey Thanks and nice post... Having one issue, my server shows avg 10 to 20 queue thread and cpu utilization 45% to 60%. doubt, since cpu is free for at least 40% then why the queue is increasing?