Performance Monitor is a utility that is used to track a variety of processes and provide a real-time graphical display of Windows 2003 system results. This tool can be used to help you with upgrade planning, monitoring of the processes to be optimized, the monitoring of the results of the tuning and configuration scenarios, and the understanding of a workload and its effect on the use of resources to identify bottlenecks.
Using Windows Perfmon to determine iSCSI performance is not the same when monitoring physical disks that are directly accessed by the system. As you know, your network setup and equipment will have the biggest impact on iSCSI performance. Having a high-performance switch that does network buffering and flow control, having good cables between the target and initiators, and having quality NICs are critical for optimal performance. ISCSI software is capable of fast speeds very close to the theoretical limit of 120MB / second. But much of the performance depends on the environment in which iSCSI is running.
You should not perform any kind of iSCSI performance analysis based solely on physical disk counters in Windows Perfmon or follow the general performance rules that many have posted on the web that relate to physical disk I / O because it is not apply to iSCSI. There is no physical correlation between the SAN physical disk queue and Windows Perfmon physical disk counters once iSCSI is involved as all disk commands are now encapsulated in network packets. Generally, you can divide the Perfmon physical disk counters by the number of disk spindles on the SAN to arrive at an approximate number of the Perfmon physical disk counters, but it is not an exact science.
A better approach for performance analysis with Windows Perfmon is to add the Microsoft iSCSI initiator classes focusing on them and the Perfmon network counters such as latency. You can also include logical disk counters, but apply the same warning as physical disk counters; they are not the only measure of iSCSI SAN performance.
PhysicalDiskAvg. The disk queue length counter indicates the average number of read and write requests that were queued for the selected disk during the sample interval. PhysicalDiskAvg. The disk read queue length counter indicates the average number of read requests that were queued for the selected disk during the sample interval. PhysicalDiskAvg. The disk write queue length counter indicates the average number of write requests that were queued for the selected disk during the sample interval.