Diagnosing Hard Disk Errors In Performance Monitor

Hard disks are probably the most annoying to diagnose using Performance Monitor, because they seem to be at the center of several different processes that interrelate with other devices. Of course, everything on a server interrelates, but hard disks are the other half of the paging duo, and that sometimes causes annoyances. Hard disks are, however, very easy to preemptively diagnose for impending failures, by simply looking at the frequency of file corruptions, and taking note of the times where the system insists on running chkdsk when the server boots.

In addition to a failing drive, disks may have problems with their i/o speed or even logic board-related problems which might point to a controller problem as well. I/O annoyances can be gauged by the Average and Current Disk Que Length (PhysicalDisk object) counters. If these counters are higher than the norm, that means that the waiting list to read and write data to and from the disk is longer than it should be. This could be a disk logic problem, or it could be a bad or improperly seated i/o cable on the drive. Worse yet, it could be a controller that’s starting to fail.

Keep in mind that Virtual Disks are available as an object too, with many of the same counters as seen here. Many of the counters that you see in the Physical Disk object generate misleading data when you’re working with any of the Advanced Volume configurations such as RAID, spanned, extended, mirrored, and striped volumes. The Virtual Disk object is ideal for these situations, and definitely should be considered when you’re trying to troubleshoot problems that may be disk controller related. Consider that a single controller might have multiple drives attached. It is sometimes easier to establish problem continuity across multiple disks, thereby allowing you to effectively eliminate the potential disk problem and focus on a controller issue.