Sams Teach Yourself MCSE Windows NT Server 4 in 14 Days
(Publisher: Macmillan Computer Publishing)
Author(s): David Schaer, et al
ISBN: 0672311283
Publication Date: 12/15/97

The following are additional memory counters of which you should be aware:

•  Available Bytes. The amount of virtual memory that can be called on by the system. Add memory if the available bytes continually remain below 4MB.

•  Committed Bytes. The amount of virtual memory that has been assigned to processes. Add memory if the committed bytes are approaching the amount of physical RAM in the system.

10.3.4. The Physical Disk

Physical disk counters show the statistics for all access to a specific physical disk. For performance reasons on older systems, physical disk tracking is disabled by default. Before tracking physical disk performance, you must execute the command DISKPERF -Y and then reboot the system. When tracking the physical disk performance on a stripe set, first execute DISKPERF -YE.

When logging disk performance, write the log file to another disk to avoid additional load on the disk, which will skew your results.

%Disk Time, as shown in Figure 10.9, tracks both read and write operations to the physical disk. Remember that a single physical disk might contain multiple logical drives.

Figure 10.9.  Note that this counter can be reported as a total or as a separate instance for each physical disk.

If the %Disk Time is constantly over 90%, it might be beneficial for you to consider moving the pagefile to a different disk, purchasing a faster disk controller, or implementing a stripe set.

Before making radical physical changes based on high physical disk readings, check the Pages/second. Remember that simply by adding more RAM you often can reduce the activity level of disk counters.

Another helpful counter is Avg. Disk Queue Length. This counter reports the average amount of disk read and write requests that are waiting during the sampling period.

10.3.5. Logical Disk

Tracking the logical disk counters shown in Figure 10.10 can be helpful when you want to determine where to place specific applications or data files. Notice that in Figure 10.10 there is a separate instance for each logical disk.

It is best to use the Free Space counter with the Alert view so that the system reports when free disk space falls below a preset level. In Figure 10.11, an alert is configured to send to NTMASTER if the free disk space on disk C falls below 100MB. If you want, an application could launch based on the alert.

Figure 10.10.  This counter can be reported as a total or per each logical disk.

Figure 10.11.  Other options include logging the event or switching to the Alert view. The counter is configured to be updated manually.

When the alert threshold is reached, an alert is generated and sent via the messenger service to NTMASTER, as shown in Figure 10.12.

Figure 10.12.  A message goes to NTMASTER in response to a previously set threshold having been exceeded.

10.3.6. The Network Segment

The Network Segment object provides information on how the system is utilizing its network media.

The %Network Utilization counter shown in Figure 10.13 represents the percentage of the media capacity in use. Although people commonly think about network implementations, such as 10BaseT, supporting 10mbps, a 10BaseT network that is performing well should run at no more than 40 percent of the segment capacity.

Although the instance enables you to select a specific network interface, this is only so that you might identify the path to the segment. The Network Segment object tracks the collective frames from all systems placed on the segment.

Figure 10.13.  The adapter is listed as an instance. If more than one adapter were present, each would be listed.

10.3.7. The Network Interface

The Network Interface object, shown in Figure 10.14, tracks information on how a specific network interface card is utilized. Although the network segment might be capable of handling additional traffic, the network interface card might be working at capacity.

Figure 10.14.  Notice that two instances are available.

By tracking the counter Bytes Total/sec you can determine whether the capacity of the network interface card is nearly full. The use of a 32-bit network interface card in the server would improve the server’s capability to handle multiple clients on the segment.

10.3.8. The Paging File

Use the Paging File object to track the utilization of the pagefiles, as shown in Figure 10.15. Because NT supports multiple pagefiles, you can specify the instance of the pagefile you want to track, or you can track the pagefiles collectively.

Figure 10.15.  Notice that the instance defaults to the individual pagefile rather than the total.

You can track either the %Usage or %Usage Peak. The first gives you a reading at the sample point, whereas the second shows the highest percentage obtained.

10.3.9. The Process Object

You should not confuse the Process object with the Processor object. The Process object lets you track the statistics of a specific process. Some services, such as the Alerter and Messenger services, do not report themselves individually but are tracked as a collective pool, as shown in Figure 10.16.

Figure 10.16.  Here you have chosen to monitor the services instance.

10.3.10. Monitoring Remote Computers

Performance Monitor enables you to monitor as many computers from your workstation as you want. The overhead of remote measurement is relatively small, but it grows with the number of computers from which you are collecting data. You can do several things to reduce this overhead.

First, lengthen the time interval between measurements. Overhead and time interval are inversely related. Next, reduce the number of objects you are monitoring. The two objects that are the most costly to monitor are the Thread and the Process objects.