Recognizing Disk Bottlenecks

Disk bottlenecks appear as sustained rates of disk activity above 85% of a sample interval and as persistent disk queues greater than 2 per disk, while paging, as measured by Memory: Page Reads/sec and Memory: Page Writes/sec, remains at less than 5 per second, on average.

High use, by itself, is a sign of demand, not a problem. In general, a high-performance disk is capable of about 40 I/O operations per second. However, nearly constant use and lengthy queues are a cause for concern. When response is poor; when you hear the disk clicking, and you see its light flashing, chart Logical Disk: Avg. Disk Queue Length and Memory: Pages/sec for all logical partitions on your workstation.

Note

Sustained high disk use and persistent long queues typically are symptoms of a memory shortage, not a disk bottleneck. When physical memory is scarce, the system starts writing the contents of memory to disk and reading in smaller chunks more frequently. The less memory you have, the more the disk is used. Rule out a memory bottleneck before investing any money in disk technology. For more information, see the following section.