C# in a Nutshell, Second Edition

Performance Counters

Event logs are useful for capturing application status that is not of a time-sensitive nature, yet needs to be recorded for future analysis. However, to gain insight into the current state of an application (or the system as a whole), a more real-time approach is needed.

The Win32 solution to this need is the performance-monitoring infrastructure, which consists of a set of performance counters that the system and applications expose, and the Microsoft Management Console (MMC) snap-ins used to monitor these counters in real time.

Performance counters are grouped into categories such as “System,” “Processor,” “.NET CLR Memory,” and so on. These categories are sometimes also referred to as “performance objects” by the GUI tools. Each category groups a related set of performance counters that monitor one aspect of the system or application. Examples of performance counters in the “.NET CLR Memory” category include “% Time in GC,” “# Bytes in All Heaps,” and “Allocated bytes/sec.”

Each category may optionally have one or more instances that can be monitored independently. For example, this is useful in the “% Processor Time” performance counter in the “Processor” category, which allows one to monitor CPU utilization. On a multiprocessor machine, this counter supports an instance for each CPU, allowing one to monitor the utilization of each CPU independently.

The following sections illustrate how to perform commonly needed tasks, such as determining which counters are ...