Some of the main topics in this chapter are
The art of monitoring the performance of Windows NT is a skill that's often overlooked until it's needed. Many people don't even run the Windows NT Performance Monitor until there's a problem. When performance problems happen, this lack of familiarity with diagnostic techniques can reduce how effective and efficient you are at finding the source of the problem. This chapter describes how to use the Performance Monitor and other tools to help diagnose your system's performance.
Finding the root of a performance problem can be challenging. Given all the variables that affect performance, try not to worry if the source of the problem is elusive. The Performance Monitor is a powerful yet fundamentally easy-to-use tool that can be a great asset. This chapter also covers some other weapons available to combat performance problems.
If you're unfamiliar with the methods and goals behind performance monitoring, spend some time now learning the hows and whys. Before explaining how the tool works, I'll give you some background on the concepts and methodology.
You might want to monitor performance for a number of reasons. The most common reason is to solve a specific performance problem. Users might be complaining about how long it takes for a certain system to respond to their requests. Performance monitoring doesn't solve the problem; it helps you understand where the problem is and how you can fix it.
For example, say that you have a Windows NT system running Microsoft SQL Server. Users run queries against the database. Over time, users begin to complain that their queries are taking longer to complete. That might be a problem with SQL Server, or it might be a basic performance problem on the system. The Windows NT Performance Monitor can help you determine the cause of the problem.
One of the possible outcomes of using the Performance Monitor is to find out whether or not there's a hardware-related performance problem. You might discover that there isn't enough memory, which is hurting the performance of the system. In that case, adding memory might resolve the problem. See Chapter 4, "Getting the Right Hardware," for more information about memory.
Using the Performance Monitor can also help you determine that the problem is with the design or use of the SQL Server database. In that case, a database administrator can change a SQL statement to improve performance. The Performance Monitor doesn't solve the problem; it lets you know where the problem is, and then you have to take the necessary steps to solve the problem.
Performance problems surprise many system administrators because they always seem to come out of the blue. Why that happens is in the very nature of performance, and you should understand why performance problems happen the way they do.
A performance problem can manifest itself in a number of ways. You might simply have poor response from a server. In certain situations, applications can fail because of performance problems--this is especially true of communications applications.
The root of the problem, however, lies in an overuse of some resource. Here are some of the resources that can cause problems:
This list obviously isn't complete, as many more resources can be a problem. This list, however, illustrates what I mean when I use the term resource.
The way a system behaves when a resource is overused varies, depending on which resource is having the problem. For example, when a processor is overused, tasks simply begin to pile up and take longer to complete. When memory is overused, however, you might see disk activity noticeably rise. The rise in disk utilization that often accompanies memory overutilization is a result of the use of virtual memory. When there's enough memory, the information stays in RAM. When there's too little memory, the Virtual Memory Manager moves the information to disk. You should check if one overused resource is affecting other resources.
The Performance Monitor refers to resources as objects. Here are some of the objects you might see listed in the Performance Monitor:
What objects, or resources, are available in the Performance Monitor varies by system. Software that you install can add objects to the Performance Monitor. For example, if you use Internet Information Server, additional objects such as HTTP Service and Internet Information Services Global will be available.
An object in Performance Monitor can have multiple instances. One example is when you're monitoring performance related to processes. Each process--and there can be many on your system--can be individually monitored. Another example is when you have multiple disk drives--there will be an instance for each physical drive as well as for physical drive performance in total. See Chapter 39, "Optimizing NT Server's Performance."
When the utilization of a resource rises, the response time of that resource rises, as well. The most common curve associated with rising use of a resource is shown in Figure 38.1.
Here is the relationship between resource utilization and response time.
At lower utilization levels, the response time slowly rises. At a certain point, often called the knee of the curve, response time begins to rise rapidly. Eventually, as the utilization continues to increase, the response time of the resource becomes unstable and unpredictable.
Notice in Figure 38.1 that no numbers are associated with the graph. The reason is that every resource hits the knee of the curve at a different utilization point. For example, processors are often considered overutilized when they reach the 70-80 percent utilization point. However, memory is sometimes considered overutilized when it reaches 30 percent utilization. Because every resource is different, I couldn't put numbers in the graph in Figure 38.1.
Now that you understand a bit about performance, jump into using the Windows NT Performance Monitor. Choose Start, Programs, Administrative Tools, Performance Monitor.
TIP: The program is named PERFMON.EXE and normally is located in the SYSTEM32 directory. Create a shortcut to it.
After you start the Performance Monitor, you see a blank screen with numbers down the left side. "Report Options and Saving the Configuration," later in this chapter, explains how to start the Performance Monitor with the information you most often want to see.
The Performance Monitor has four modes, or views:
I normally use Chart mode when I'm focused on one system and I'm trying to get an overview of the system's performance. Because I don't see the underlying data, I don't use the information to determine what's wrong. Instead, I use it to narrow down what I need to look at further.
I use Alert mode as a monitor. I often start Alert mode on a system running Windows NT Workstation and use it to monitor multiple systems running Windows NT Server. (See how to do this in the section, "Using Alert Mode.")
I use Log mode when I'm trying to reproduce a problem. For example, if I run a specific SQL Server query and the performance is quite bad, I create a log. Because I can't watch the Performance Monitor and run the query. Instead, I can use the log and review the data after I've run the test.
Finally, I use Report mode to look at the numbers, especially in a side-by-side format. For example, if I have multiple physical or logical drives, I can watch the performance numbers for the drives next to each other, so I use Report mode as another method for narrowing down a problem to a specific drive.
Normally, the Performance Monitor starts in Chart mode. If you find yourself in another mode and want to switch to Chart mode, there are several ways to do this. Figure 38.2 shows the toolbar for the Performance Monitor. The button on the far left side, selected in the figure, takes you to Chart mode.
Here, the Performance Monitor toolbar is in Chart mode.
You can also choose View, Chart or press Ctrl+C to access Chart mode.
Use the Add to Chart dialog box (see Figure 38.3) to add objects that you want to chart. You bring up this dialog box by clicking the plus sign on the toolbar; choosing Edit, Add to Chart; or pressing Ctrl+I.
Starting from the top, the first item you see is a place to enter a computer name. By clicking the button labeled with the ellipsis on the right side of that text entry field, Performance Monitor shows a list of systems from which you can select. The important point to understand is that you can have data from multiple systems on the same chart. This is true of the other modes, as well.
The objects listed in the Object drop-down list box vary, based on what software you have installed.
The next field is the drop-down list box that contains the list of objects you can monitor. As I said earlier, different objects appear in this list box depending on what software you have installed. If you select a different system, different objects can be in this list box compared to your local system. The object that you select affects what appears in the next field.
The Counter field lists the specific data that you can chart for this object. If you want an explanation of what a specific counter means, click that counter and click the Explain button. This adds a field to the dialog box with a description of what the counter means.
At the bottom of the dialog box are four places to customize the way the data is plotted on the chart. You can change the color of the line, set a specific scale for the counter, set a line width, and choose a line type. Later, I describe how to change these options after you've already selected a counter and have the chart running.
To add a counter to the chart, click the Add button. When you're satisfied with the counters you have displayed, click Cancel. (The button changes to Done after you've selected at least one counter.)
In Figure 38.4, I've selected counters for Processor, Physical Disk, and Memory Utilization. The counters are listed along with the line color for that counter. Above that list is statistical information for the counter currently selected; this is called the value bar. In the value bar in the dialog box, you see the last value, average, minimum, maximum, and graph time for % Disk Time. If you click Page Faults/sec, you see the statistical data for that counter.
If you decide that you're unhappy with the way the chart looks, you can change the graph without removing and adding the counter. If you double-click the counter at the bottom of the window, the Edit Chart Line dialog box appears, enabling you to change the color, line type, width, and scale.
If you decide to remove a counter, click the counter and press Delete. There are buttons on the toolbar to edit and delete a chart line, along with other charting options that you can access by clicking the button on the right side of the toolbar; choosing Options, Chart; or pressing Ctrl+O. This brings you to the Chart Options dialog box (see Figure 38.5).
The Performance Monitor can monitor processor, disk, and memory utilization in real time.
In the Chart Options dialog box, you can modify the general characteristics of the chart you want to display.
In the Chart Options dialog box, you can select whether the legend, value bar, vertical and horizontal gridlines, and vertical labels are displayed. You also can choose whether to display the chart as a line chart or histogram. You also can specify that the data automatically updates at certain intervals--and what that interval should be--or that you want to choose when the data is updated.
If you select Manual Update in the Chart Options dialog box, you have to force the Performance Monitor to update the data. You can do this by clicking the Camera icon on the toolbar or from the Options menu. With the Manual Update option, you can reduce the load placed on a system due to automatic monitoring updates.
After you've configured the chart the way you want, might want to save that configuration. It's a bit of a pain to have to go back in and select all the counters and options over again, so there's a way to save these so that you can quickly pull them up again.
Choose File, Save Chart Settings As to save your chart configuration. The extension used for a chart settings file is .pmc. After you have saved a chart settings file, choose File, Open to use those chart settings. If your chart has preselected counters and uses periodic updates, your chart will immediately start displaying live data.
That's how you save the chart configuration, but what about the chart data? You can choose File, Export Chart to export the data in either tab- or comma-delimited format. Figure 38.6 shows an example of chart data imported into Excel.
Chart data is exported from the Performance Monitor and imported into Excel by using comma-delimited format.
I use Alert mode more often than the other modes. My experience with a system tells me what threshold different resources can handle before they're overcommitted, so I set alerts for the same component on different systems at different levels. Those alerts then let me know how often that component is overused.
Alert mode also can write the alerts to the application event log. That way, you have a single place to look for problems with a system.
To access Alert mode, do one of the following: click the second button from the left on the toolbar, choose View, Alert, or press Ctrl+A. Figure 38.7 shows the basic Alert window.
The Alert window in the Performance Monitor is used to track when resources reach a pre-specified utilization threshold.
The process of selecting which resources, or objects, to monitor in Alert mode is similar to the process used in Chart mode. Probably the easiest way to start is to click the plus button on the toolbar. You see the Add to Alert dialog box shown in Figure 38.8.
In the Add to Alert dialog box, you specify the details of the alerts you want to monitor.
The Computer, Object, Counter, and Instance fields are used the same way they were used in Chart mode. Some additional fields are available in this dialog box, however, that weren't available in Chart mode. Those fields are in the Alert If and Run Program on Alert frames.
The Alert If frame enables you to enter a specific value that, when reached, triggers the Performance Monitor to tell you that the threshold was hit. You can enter a minimum or maximum value. What you enter depends on the counter that you've selected. If you select a counter that provides data in percent, such as % Processor Time, then you enter a value between 0 and 100. If you select a standard value, such as Page Faults/sec, enter the number at which you want to be notified.
NOTE: The number you enter as the threshold relates to the interval you specify for Alert mode. The default interval is five seconds. If you enter 75 percent for the Processor Time as the alert, the system must average 75 percent utilization for five seconds before triggering the alert. You won't be notified if the system peaks once just momentarily at 75 percent. In the following section, you learn how to change the interval for Alert mode.
The Run Program on Alert frame enables you to have a program run when your threshold value is reached. The program can run the first time the threshold is reached or every time the threshold is reached. You can use a batch file or a program, and the alert information is passed as a parameter to the batch file or program.
Just like Chart mode, you can select multiple counters for Alert mode by clicking Add. When you're finished, click Cancel, which will read Done after you've added at least one counter. You can see what the Alert window looks like in use in Figure 38.9.
When using Alert mode, several counters are selected including % Processor Time, Page Faults/sec, and % Disk Time.
At the bottom of the window, notice the Alert Legend. This shows you what counters you have alerts set up for, what values trigger the alert, and what system the counter is for. You can delete an alert by clicking it in the Alert Legend box and then pressing the Delete key or clicking the Delete button on the toolbar.
If you're not satisfied with the way you have alerts set up, you can change them. Double-click an alert in the Alert Legend box to see the Edit Alert Entry dialog box. Here, you can change the color associated with the alert, the threshold value for the alert, and the program that's run as a result of reaching the threshold value. You can also choose Edit, Edit Alert Entry.
Another way to customize Alert mode is to click the button on the right side of the toolbar. This brings up the Alert Options dialog box, which you can see in Figure 38.10. You can also choose Options, Alert or press Ctrl+O.
Use the Alert Options dialog box to change the overall characteristics of Alert mode.
Use the Switch to Alert View check box to have the Performance Monitor automatically show you the Alert window whenever an alert occurs. For example, if you're monitoring system performance in Chart mode but also have alerts set up, Alert mode pops up any time one of the threshold values you've specified is reached.
The Log Event in the Application Log check box causes the Performance Monitor to write an entry into the Application Event Log every time one of your threshold values is reached. This is a handy feature because you can keep the event log for later review.
The next item in the Alert Options dialog box is the Network Alert frame. This frame is used if you want to send a message to another system when your threshold value is reached. When you select the Send Network Message check box, the Net Name Entry field is enabled so that you can enter the name of a computer on your network. You need to have the messenger service started before you can use this option.
The Update Time frame is just like the one in Chart mode. You can have alerts checked for specific predefined intervals, or you can manually force the Performance Monitor to check by clicking the Camera icon on the toolbar.
After you create a set of alerts, you might want to monitor those alerts on an ongoing basis. The easiest way to do this is to save the alert configuration that you created. You can do this by choosing File, Save Alert Settings As and saving your settings into a file with an extension .pma.
You then can choose File, Open to load your settings. If you have alerts set up with periodic updates, the process will start automatically. You can also save the alert data by choosing File, Save.
Like Chart mode, you export the data into either tab- or comma-separated values. Choose File, Export Alert to accomplish this. Then you can import the data into programs that support tab- or comma-separated files, such as Excel.
There are a couple of situations in which I commonly use Log mode. One is to analyze performance while I'm doing something on a system. For example, I start the logging of performance data, run a SQL Server query, and then review the performance of the system based on the logged data. This happens quite often.
I also use Log mode to collect long-term performance data. I use the exported data I collect on a regular basis for capacity-planning purposes. The logged data shows me utilization trends that I can graph and present to management. That way, there's some warning as to when systems need to be upgraded.
The size of the log file varies widely. It depends on the data being collected and the frequency of the data collection.
To access Log mode, click the third button from the left in the toolbar; choose View, Log; or press Ctrl+L. You see the Log Mode window shown in Figure 38.11.
The first point to notice on this screen is the field on the right side labeled Status. Notice that it says Closed, which tells you that no logging is taking place at this time. When you start a log, the status changes.
To start logging performance data, click the plus sign button (or choose Edit, Add to Log; or press Ctrl+I). When you do, you see a dialog box similar to the one shown in Figure 38.12. This is where you specify what objects you want to start logging and what system those objects are on.
The Computer field is used to select from which system you want to collect data. You can collect data from multiple systems in a single log file. If you click the button with three dots on the right, the Performance Monitor displays a list of systems from which you can select.
The Log Mode window in the Performance Monitor stores detailed performance data to a file for later analysis.
Select objects that you want to log. No counters are listed here, as in Chart and Alert modes because Log mode stores all counters for an object.
In the Object field, you select which objects, or resources, you want to collect data for. With Chart and Alert modes, there were counters for each object. You don't see counters listed here, however, because the Performance Monitor logs all the counters for an object you select. You get very detailed data when using Log mode.
Select multiple objects by clicking them and then clicking Add. In Figure 38.13, you can see that I have several objects selected for logging, yet the Status field shows that logging hasn't started yet.
Another step is necessary to start the logging process: You need to bring up the Log Options dialog box (see Figure 38.14). Do so by clicking the button on the right side of the toolbar; choosing Options, Log; or pressing Ctrl+O. You need to specify the directory and file name for the log file before you can start logging.
You need to specify the log file before logging begins.
In the Log Options dialog box, you tell the Performance Monitor the name of the log file and start the logging process.
Also in the Log Options dialog box is a Start Log button. Like Chart and Alert modes, Log Mode offers you the option at the bottom of the dialog box to have the data logged at regular intervals or to manually update the data. If you opt to manually update the data, you'll click the camera button on the toolbar to force an update.
After you specify the name of the log file and click Start Log, the file is created and logging begins. The status changes to Collecting. The Performance Monitor also puts the log file name in the Log File field and shows the size of the log in the File Size field. The file size increases over time as entries are added to the log.
After you collect all the data you need, you can stop the log. To do so, bring up the Log Options dialog box again. As you can see in Figure 38.15, the button previously labeled Start Log is now labeled Stop Log. When you click this button, logging of the data stops, and all the data is available in the log file.
When you're ready to stop the log, bring up this dialog box and click the Stop Log button.
After you've collected log data, you can use that data in the other Performance Monitor modes. I'll only walk you through the process of using log data in Alert mode because using log data in other modes is similar.
First, switch to Alert mode. The Performance Monitor assumes that you want to use current data, not data from a log. To use data from a log, choose Options, Data From. Use the dialog box that appears to tell the Performance Monitor that you want to use data from a performance log and the name of that log. Click the button with the three dots to search for the log file.
After you've selected a log file, the Performance Monitor behaves a bit differently. You still go through the normal process of selecting computers, objects, and counters to show alerts for. However, the list of available computers, objects, and counters is now based on the data in the log file. For example, if you've collected data from only one system, you will see only the alerts for one system. If you've only collected data about processor utilization, you will see only alerts related to processor utilization.
Four objects are available in the drop-down list box: IP, Memory, PhysicalDisk, and Processor. Those were the only four objects I selected when I collected the log data. What is available here correlates to what I selected in Figure 38.13.
For the four objects for which I have data, all the counters are available. I select any or all of the counters and add alerts just as if the data were live. Notice on the window that the Run Program on Alert frame is disabled. Because the data isn't live, it doesn't make sense to run a program when an alert happens.
After I select all the desired counters, all the relevant alerts in the data are shown (see Figure 38.16). About the only difference between this view and the view you saw in the earlier section on Alert mode is in the status bar at the bottom of the window, which now indicates that the data comes from a log file.
You can also change the time frame that's shown by choosing Edit, Time Window or pressing Ctrl+E. The Input Log File Timeframe dialog box appears, showing the time range of the data in the log file. You can move around the sliding scale to adjust the time frame used to display the data. Bookmarks (discussed shortly) are also shown.
These alerts are shown, based on data from a log file.
You can export the data from the log when you collect it. To do so, choose File, Export Log. Once again, you can save the data in tab- or comma-delimited format.
TIP: Make sure that you export the data as soon as you collect it. If you open the log file later, you can't export the data then. If you think you will need to use the data in other applications, export it just to be safe.
The bookmark feature enables you to add placeholders in the data wherever you want. Locate the button on the toolbar that looks like an open book with the corner of a page turned down. Whenever you click that button while the log is collecting data, the Performance Monitor adds a bookmark.
I use bookmarks to let me know when I took certain steps. For example, if I'm evaluating the performance of a SQL Server query, I use a bookmark to note when I start the query, the phases the query goes through during processing, and when the query finishes. These bookmarks make interpreting the data much easier.
You can save your log settings by choosing File, Save Log Settings As. The log settings are saved to a file with the extension .pml. The next time you want to use your log settings, choose File, Open and select your log settings file.
Report mode behaves much like Chart and Alert modes, with the key difference being that Report mode shows you the raw numbers as they're happening. Report mode is also handy when you're trying to compare the performance of two systems. The layout of the data makes that easy to do.
Access Report mode by clicking the fourth button from the left on the toolbar; choosing View, Report; or pressing Ctrl+R. When you start Report mode, you see a blank white screen with the usual cast of suspects on the toolbar.
To add objects to a report, click the plus sign button on the toolbar (or choose Edit, Add to Report; or press Ctrl+I). You see a dialog box resembling Figure 38.17.
The Add to Report dialog box is used exactly like the Add to Chart dialog box.
Notice that this dialog box looks almost identical to the dialog box used in Chart mode. If you took the time to understand the Add to Chart dialog box shown in Figure 38.3, you already know how to use the Add to Report dialog box. The only difference is that because you aren't charting the data, you don't have options related to the color of the line, size and type of the line, and scale.
Select which computers you want data from in the Computer field. You select resources that you want data from in the Object drop-down list box. Select the counters you want to report on in the scroll box, based on which object you selected. Click the Add button to add the counter to the report.
If you're displaying live data, after you've selected the counters to add and have clicked the Done button, the data displays. Figure 38.18 shows an example of a report that includes processor time, page faults, and information about the utilization of the logical disk drive.
Like the other modes in the Performance Monitor, the information in Report mode is averaged over the time interval used in the collection. Most of the information on this report is pretty easy to understand, one of the benefits of the layout of this report being how easy it is to read.
My hard drive is partitioned into drive C and drive D. This report shows the utilization of the two logical drives side-by-side.
In Figure 38.18, you can see the utilization of the two drives side-by-side, making it easy to compare how different resources are being loaded. Seeing one resource in a set of like resources more heavily loaded than the others might encourage you to do some reconfiguring. If the two resources are hard drives, you might move some of the data to the underutilized drive to even the load.
You can change the collection interval or make the data collection process manual by clicking the button on the right side of the toolbar. You can also choose Options, Report, which brings up a dialog box, enabling you to change how often the data is updated. Keep in mind that the data shown is an average over the collection interval. The default collection interval for a report is five seconds.
Like the other modes, after you get a report configuration that you want, you can save that configuration. Choose File, Save Report Settings As to save your configuration into a file with the .pmr extension. You can use that configuration file later by choosing File, Open.
You can also export the report to a file with a tab- or comma-separated format. Choose File, Export Report to accomplish this. The report exports well and looks just like what you see on-screen when you pull it up in a program such as Excel.
Running the Performance Monitor on a system can put a noticeable load on the system's resources. The amount of the load varies, depending on what you monitor and what you do with the data. For example, graphing network utilization can take quite a bit of CPU power. Try to be careful and not be the process that puts a heavily loaded system into overload.
One way to avoid overload is to use the Performance Monitor on a Windows NT Workstation system to monitor your servers. There is a bit of a trade-off involved in this, as monitoring a remote system increases network traffic. In a situation in which you're monitoring network utilization, therefore, you might want to do it from the server you need to watch.
You might have noticed how I displayed disk utilization information in some of the examples. There's a trick involved in getting the Performance Monitor to show disk performance data. People who haven't used the Performance Monitor much sometimes become frustrated because all the disk utilization data stays at zero.
You need to run the DISKPERF command with the -y or -n switch to cause the system to collect disk performance data or not. If you have striped disk sets, you need to specify -ye to collect disk performance data.
For each of the four modes, I discussed saving the configuration that you create so that you can easily see the same sets of data again. There's another way, however, to save the configuration.
Saving the way all four modes are configured is called saving a workspace. A workspace file has the .pmw extension. The File menu in all four modes contains a Save Workspace menu item, so whenever you tweak your Performance Monitor configuration just right, you can save the entire configuration.
You can start the Performance Monitor with a specific configuration file. From a DOS prompt, run the PERFMON.EXE command followed by the name of a configuration file, and that configuration file automatically loads.
You can even specify the name of a default system other than the one on which you're running. In the PERFMON.EXE command, just include the UNC name of a system in your network:
perfmon c:\temp\mywkspc.pmw \\SYSNAME
This command runs the Performance Monitor by using the workspace named mywkspc and a default system of SYSNAME. You see the default system in all the dialog boxes in which you're able to select a computer. For example, in Figure 38.12, SYSNAME would show up in the Computer field. Instead of the workspace file, this example could have used a file with an extension of .pmc, .pma, .pml, or .pmr.
One of my favorite tricks with the Performance Monitor is to remove all the toolbars, legends, menu bars, and so forth, making a clean window for viewing a chart. Then I decrease the size of the window and place it toward the bottom or top of my desktop. This makes the Performance Monitor much less obtrusive during day-to-day use.
You can create a chart like this by first removing all the legends and vertical labels in the Chart Options dialog box. Then, from the Options menu, remove the check marks next to Status Bar, Toolbar, and Menu and Title. You now can size the window any way you want.
Another tool included in Windows NT aids in monitoring performance: Task Manager. The new version 4.0 shows a great deal of performance information about individual processes and the system as a whole. I often turn to Task Manager first to give me some clues about what's causing a performance problem on a system.
One way to run Task Manager is to press Ctrl+Alt+Delete. In Windows NT, this combination brings up the Security window; Task Manager is one of the buttons on the Security window. You also can run the program TASKMGR.EXE or right-click the taskbar and choose Task Manager.
The Performance tab of Task Manager has a scrolling graph that shows CPU utilization and memory utilization. There's also information about threads, physical memory, commit charge, and kernel memory.
The status bar gives additional information about the number of processes, CPU utilization, and memory utilization. The Processes tab shows information about the activity of individual programs and processes. Included on this tab is a wide variety of performance information. It enables you to find the details concerning how specific processes are using system resources, and thus is a quick way to find the processes that are using the most resources on your system.
Clicking the title of a column sorts the data by that column. If I'm trying to find the process with the highest level of CPU utilization, I can sort by that column. Other columns I often use include Total CPU Time Used and Memory Usage. Here's a complete list of the columns available:
If you're experiencing performance problems, start by using Task Manager. It helps you understand whether you're dealing with a process or a system that has a performance problem.
© Copyright, Macmillan Computer Publishing. All rights reserved.