Windows 2000 Server Performance Monitor is used to monitor several important performance counters to determine if the load placed on Small Business Server 2000 is causing bottlenecks in any critical hardware resources. The primary resources tracked include memory, processor, and disk subsystem.

The best indicator of a memory bottleneck is a sustained, high rate of hard page faults. Hard page faults occur when the data a program needs is not found in its working set (the physical memory visible to the program) or elsewhere in physical memory, and must be retrieved from disk.
An acceptable range of page faults per second is 0 to 20. The bottleneck stage of memory page faults per second is over 50.

The cases that are normally close to a bottleneck stage were the 256MB RAM tests, which were consistently averaging close to 28 pages per second. Although this indicates fairly heavy use of the page file, memory usage is not a bottleneck.

Processor activity is especially important for server-based applications such as SQL Server and Exchange. Two of the most common causes of CPU bottlenecks are CPU-bound applications, and excessive interrupts that are generated by inadequate disk or network subsystem components.

Processor use in excess of 75 percent is considered a bottleneck.

Disk performance counters include the percent of disk time and the disk queue length. In this test, the average percentage of disk time was normal, about 45 percent when all load simulators were active. Anything over 2 is considered a bottleneck.

It should also be noted that the Performance Monitor performs more page faults than all of the other processes combined, which more than doubled the page file load on the disk subsystem. Recommended improvements would include using a hardware RAID controller rather than software RAID, and moving the page file off of the system partition.

You might also consider adding more physical disks to the SCSI or hardware RAID array because each disk spindle (disk arm) will take on additional file I/O traffic. More, smaller disks will be a better solution than a few larger disks.