Chapter 4
Introducing The Performance Monitor

•  Creating Charts

•  Creating Logs

•  Creating Reports

•  Creating Alerts

The Performance Monitor is used to isolate performance bottlenecks. The Performance Monitor can also be used to help you tune the additional Microsoft BackOffice components, so gaining a basic understanding of its usage is quite important. However, if you find that the Performance Monitor is just not your cup of tea for performance tuning, it can still be useful for generating alerts. An alert is a means of notifying you of a problem, such as when a server is short of disk space.

We will start our discussion with the basic functions of the Performance Monitor. This includes how to create and configure charts, logs, reports, and alerts. I want you to keep in mind that this subject can be very complex (performance tuning with the Performance Monitor is an art—not an exact science), and there are books devoted to just this particular program and its usage. So, don’t be concerned if it takes you some time to really get the hang of using the Performance Monitor and its associated performance counters.

Using The Performance Monitor

The Performance Monitor is used to monitor a system in real time. By this, I mean that the event objects you monitor are occurring right now, and the value you see for the event object reflects the actual value of the event object with a minimal time lag. You can use the Performance Monitor on the computer you want to monitor, which will affect the performance on that computer slightly. Or, you can use the Performance Monitor remotely from another computer, which will impact your network bandwidth more than it will affect the performance of the computer you are monitoring. The amount of performance degradation will vary from unnoticeable to appreciable depending on the number and frequency of monitored events.

The Performance Monitor defines events, as I put it, as objects. An object can be a system, processor, process, thread, or similar item. Objects are further divided into counters. For example, the system object includes counters for % Total Processor Time (the total amount of processor usage on the system), % Total Interrupt Time (the total number of interrupts generated on the system), % Total Privileged Time (the total amount of processor time spent in the kernel on the system), among others. And within a counter, you may also have one or more instances. As an example, if you have more than one processor on your system and view the Processor Counter % Processor Time (the total amount of processor utilization for a particular processor), then you will find multiple entries in the Instance field. Instance counts start with zero and increase by one for each additional item. If you have two processors, then your Instance field will include 0 for the first processor and 1 for the second processor. You could also have multiple named instances rather than numbers. If you monitor the RAS Port object counter, for example, the Instance field could include COM1, COM2, COM3, and so on, up to as many RAS connections as you have installed.

The Performance Monitor supports four types of views, each of which displays the performance object counters in a different format to provide you with unique capabilities. These are:

•  Chart—This view can display values for performance counters in either a line (graph) or bar (histogram) chart. A line chart is useful for quickly displaying performance data over a period of time, and a histogram chart is useful for displaying performance peaks. Be careful, though, because a chart can become cluttered and difficult to read if you attempt to view more than 10 active performance counters at one time.

•  Log—This view is used to capture the performance counters for a particular object to a file. This captured file may be read into the Performance Monitor at a later date for in-depth analysis. You can load a log file and then chart it, for example.

---

TIP:  When using the Log option, you should remember that you can only view object counters that you have previously captured. However, you do not have to view all of your captured object counters. You may view a subset of your counters. This is one way to start with the big picture and then narrow it down to a specific incident. So, it is always better to capture any performance counters that you think you may need, rather than limit the capture to a few specific performance counters that you know you will need.

---

•  Report—This view is used to display a large series of performance object counter values in real time. It is an instantaneous view of the performance counters, so you can only view the current value. This value will change at the next update interval, which means the value you are monitoring may be reset to zero before you can evaluate it.

•  Alert—This view is used to set conditions based on performance object counters, and it alerts you when the condition is met. For example, you may want to be informed when your server’s paging file grows beyond its current minimum allocation. This way, you can increase the minimum allocation size and increase overall system performance. Anytime a paging file grows beyond the minimum, there is a significant amount of overhead involved in expanding the page file. Additionally, any page file expansion may be fragmented, which will slow down the retrieval of paged data.

In the next few sections you will explore how to create charts, logs, reports, and alerts. But before we get into how to create these different views, a brief explanation of the Performance Monitor Toolbar is in order.