Q We are a small business running Microsoft NT Server 4.0 on a DEC Prioris XL 5166 server (130MB RAM). Recently we have noted that the server is getting slower and slower. Once in a while the network administrator shuts the server down and restarts it; it runs fine for a few days and then starts to slow down again.

A When an application starts it requests resources, particularly system memory. Often, when the application performs various operations it requests additional system resources. What is most likely happening to your server is that there is some software that fails to release the system resources when it is finished with them.

To get a closer look at what is happening on an NT computer, run Performance Monitor (perfmon.exe). To select the resources you want to monitor, choose Add to Chart from the Edit menu. From the Object section of the dialogue box, choose Memory. From Counter choose Page faults/sec. Click Add to start monitoring these items. When you are using virtual memory, some of the items in "memory" are in the computer's physical RAM and some are "paged off" onto the hard drive. Whenever you attempt to access something that has been paged off onto the hard drive, this is known as a "page fault". Regular page faults are an indication that you do not have sufficient RAM in your computer ù look for an increase in the number of page faults over time.

Caption: For an under-the-bonnet view of Windows NT, use the Performance Monitor

To monitor the amount of free memory, choose Memory from the Object section and Available bytes from Counter. If you have a problem application, then the number of Available bytes will decrease over time.

Another thing to look at is the number of threads in a system. Operating systems such as Windows 95, 98 and NT are multi-threaded ù that is, applications can be broken into sub-applications called threads. To monitor threads, choose Objects from Object. From Counter, choose Threads. Click the Add button. Sometimes there will be faulty threads that start up and obtain system resources, but when they have finished executing, fail to close. You will see this as a continual increase in the number of threads.

Because your problem builds up over several days, it important to log the data so you can analyse it later. To create a log file, go to the View menu and click Log. Now go to the Edit menu and choose Add to log. From the Objects section choose which object you want to log, such as Memory or Thread. Click on Done to close the dialogue box. Now go to the Options menu and choose Log. Enter a filename. To start logging, click Start Log.

To view the data from a log, choose Data From from the Options menu. Click the Log File radio button and type in the name and path of the log file.

So far I have described how to determine whether a particular type of problem is occurring. Finding a solution is more difficult! However, the first step would be to temporarily disable (or upgrade) server applications and see if the situation improves.