January 9, 2009

The best way to know when a problem is brewing is to know how things perform when all’s well with the system. You need to establish a baseline – a static picture of your network and servers when they are working correctly. One of the common tools used to create a baseline is the Performance Monitor utility that comes with Windows NT/2000/XP (but you can also create baselines using most network management utilities).


Administrators use Performance Monitor (PerfMon) to view the behavior of hardware and other resources on NT/2000/XP machines, either locally or remotely. PerfMon can monitor both real-time and historical data about the performance of your systems. To access the Performance Monitor applet, choose Start/Programs/Administrative Tools/Performance Monitor from any Windows NT machine. Windows 2000/XP machines call the option simply “Performance.”
Once you access Performance Monitor, you need to configure it to display data. The process of configuring Performance Monitor requires you to understand the concept of objects, counters and views. An object in Performance Monitor relates directly to the component of your system that you want to monitor, such as the processor or memory. Each object has different measurable aspects, called counters. Counters, in other words, are the portions of an object that you want to track. as you decide which object(s) to monitor in your system, select one or multiple counters for each object. Add these counters to whichever view you need to use. Performance Monitor can display selected counter information in a variety of views, with each view imparting different types of information. The Log view, for example, lets you store data about your systems to be reviewed later. This is the view used to create a baseline, although the other views (i.e. Chart, Alert, and Report) are useful for troubleshooting problems as they arise.
To access the Log view, either click the Log view button or choose view/Log. To add objects to the Log view, either click the Add To button (the + sign) or choose Edit/Add To Log. In the Add To dialog box, first select the computer to monitor. Choose either the local machine (the default) or a remote machine. To monitor a remote machine, type in the computer name using the Universal Naming Convention (UNC). To monitor a machine named HOUBDC1, for example, you would type \\HOUBDC1 in the computer field. You can also use the Select Computer button (at the right end of the Computer field) to view the available machines and select the one you want to monitor.
While it is often easiest to monitor a machine locally, it is often more accurate to monitor the machines remotely. Performance Monitor running on a machine uses a certain amount of resources to take the measurements and to display data graphically. Especially when you troubleshoot issues with disk performance, memory and paging, or processor use, you should not corrupt your results by monitoring locally. There are some cases where monitoring locally is preferred or required. If you are monitoring network access or networking protocol objects, for example, monitoring locally will affect the readings less than monitoring remotely. Similarly, you must monitor a system locally if you cannot access the system over the network. Finally, when you monitor objects created by a specific application, such as Exchange, you should monitor locally, as the objects related to this application are only created locally and will not be available from another system.
Once you have selected a system to monitor, either locally or remotely, you must select the object to monitor. Select one or more objects to monitor from the list in the Object Field. Note that the Log view is somewhat different from the other views in that you only add objects to the view, not the specific counters for objects.
After you select the objects for Performance Monitor to track and log, select Options/Log Options to save the data to a log file and to start the logging by clicking the Start Log button. The dialog box also gives you the opportunity to select the update method and time.
After you have configured the log to save to a particular file, you can see the log file name, status of the logging process, log interval, and file size of the log in the Performance Monitor dialog box. To stop collecting data in a log, open the Log Options dialog box again and click Stop Log. You can then choose to create a new log file and begin logging again, if necessary. You will also have the ability to view data from one of these saved log files by selecting Options/Data From. In the Data From dialog box, you can choose to continue obtaining data from the current activity or to obtain data from a particular log file.
When you choose to obtain data from a saved log, you go back to that frozen moment in time and add counters to the other views for the objects you chose to save in the log. You may want to select a wide variety of objects, so that when you open the log to display in any of the other views (i.e. Chart, Alert and Report), you can add any counters necessary.

NetWare Monitor

On a NetWare server, most of the critical information you might need to see and document to establish your baseline can be obtained by loading the Monitor application on the server itself (you can view the program remotely on a client PC, but it runs on the server). Novell calls a program that runs on the server in this way a NetWare Loadable Module or NLM, and you issue the command ‘LOAD MONITOR’ at the server’s console prompt to start the program.
The Monitor NLM can display a wide range of information, from memory usage to individual statistics about the NIC’s installed in the server. Many system managers leave Monitor running all the time so that they can keep an eye on things; it can also be used to kick users off the server and see which files they are accessing.

%d bloggers like this: