Analyzing Processor Activity

Uniprocessor monitoring and analysis involve many variables. The following steps summarize in-depth monitoring and analysis of processor activity.

• Establish a baseline for processor performance that reflects your systems typical workload. Perform this step to characterize your systems workload and identify how applications use the system.

• Examine overall processor usage by viewing processor queue length and processor utilization (also referred to as processor time). Perform this step to obtain an overview of how heavily your system uses its resources.

• Examine activity that adds to the processing load, such as high rates of interrupts and context switches. Perform this step to determine the efficiency of your system.

• Examine individual processes and their percentage of the overall processor time. Perform this step to learn about the processes on your system.

• Examine the threads— the units of work that make up a process—for each individual process and each threads processor usage. Perform this step to learn about thread utilization on the system.

• Evaluate thread priorities and change them to see if this provides better performance. (Microsoft does not recommend this as a long-term solution, but suggests it for testing purposes.) Perform this step to learn how threads interact and to determine which threads are preempting other lower-priority threads.

Figure 29.1 illustrates the role of processor monitoring in overall system monitoring.

Figure 29.1 Role of Processor Monitoring in Overall Monitoring Sequence

Before you begin the monitoring process, become familiar with the counters designed to measure processor activity and the Windows 2000 Resource Kit tools that can give you more information about processor workload and performance. The following sections summarize these counters and tools.