Comparing current performance with a previously established baseline is an important step in tuning a Windows 2000 disk system. This is especially important in a heterogeneous storage environment
Windows 2000 provides a set of tools for measuring disk efficiencies in its System Monitor, including Disk Read in bytes/sec and disk reads and writers per sec and disk queue length. However, in order to get a truly comparable statistics, Microsoft recommends the following steps when testing disk efficiencies.
1. Defragment your disk. Disk fragmentation can have a major impact on disk performance by itself and it tends to increase seek time and I/O writes, which can distort your measurements.
2. Make sure the disks being tested are not compressed or encrypted unless you use those features. Both compression and add overhead. If you are using encryption or compression, make sure to collect your baseline data with encryption or compression turned on.
3. Log performance data to another physical disk or computer. The very act of writing performance statistics can significantly slow disk throughput and distort the results of tests. Ideally, results should be logged to another computer to minimize performance effects. If that isn't possible, Microsoft recommends logging the results to another physical disk on the system. Third best is to log to another logical volume on the drive. Finally, you can measure the disk monitoring overhead during an idle period and subtract that figure from your performance data.
4. As much as possible work with individual instances. Using the _Total_instance and Summed values providing by Windows 2000's reporting tools can overstate the results.
For more information:Tip: Fast guide to Windows storage
Tip: Simplify disk alignment in Windows 2000
Tip: Nuances of Windows NT and SCSI disk performance
About the author: Rick Cook has been writing about mass storage since the days when the term meant an 80K floppy disk. The computers he learned on used ferrite cores and magnetic drums. For the last twenty years he has been a freelance writer specializing in storage and other computer issues..
This was first published in May 2004