The symptoms of a processor bottleneck aren't difficult to recognize:
But these symptoms don't always indicate a processor problem. And even when the processor is the problem, adding extra processors doesn't always solve it. In this chapter, you'll learn to use Performance Monitor to analyze such symptoms, determine the likely cause of processor bottlenecks, and implement effective solutions.
Note
Before upgrading or adding processors, verify that the processor is the source of problem. Memory shortages, by far the most common bottleneck, often masquerade as high processor use. For more information see Chapter 12, "Detecting Memory Bottlenecks."
For more information on monitoring processor use on multiprocessor computers, see Chapter 16, "Monitoring Multiprocessor Computers."
Use the following counters to measure different aspects of processor use.
Object | Counter | |
System | % Total Processor Time | |
System | Processor Queue Length | |
Processor | % Processor Time | |
Process | % Processor Time | |
Process | % User Time | |
Process | % Privileged Time | |
Process | Priority Base | |
Thread | Thread State | |
Thread | Priority Base | |
Thread | Priority Current | |
Thread | Context Switches/Sec | |
Thread | % User Time | |
Thread | % Privileged Time |
It is also useful to log Memory: Pages/sec, Logical Disk: % Disk Time and an activity count for your network to rule out problems in these components.