This was the appropriate step for his particular application, but it might not make such a difference with the ones you run. How do you know where the bottlenecks in your system lie?
Windows 2000 Professional is able to provide you with extensive performance metrics to help you locate the system snarl-ups. Unfortunately, the process itself is somewhat arcane and riddled with cryptic terms.
The tool to use is the Performance Monitor Management Console. It's located under Start-Programs-Admini-strative Tools-Performance, or run it by simply typing in perform in the Run dialogue.
First, set up a log file. Expand the branch under Performance Logs and Alerts and select Counter Logs. Either right-click in the right-hand pane or use the Action menu to add a new log file with the New Log Settings command; the log file will be stored in the \Perflogs directory of your system drive.
The log file format can be in binary System Monitor format, or Comma Separated Values (good for viewing the log in Excel, for example) or a plain text file. You can set Performance Monitor to stamp the date and time automatically on the file name as well, and decide when to run it and at what intervals to collect samples.
Performance Monitor offers a large number of "performance objects" and counters from which to select. If you want to check your hard disks, you need to take one extra step: open a cmd box, and type diskperf -y.
You'll see the message "Both Logical and Physical Disk Performance counters on this system are currently set to start at boot", so it's time to reboot.
Here are some of Microsoft's suggestions from the Knowledge Base for finding out if you need to throw more memory, CPU horsepower or a quicker/larger disk at your system:
MEMORY Pick the Memory object and add the Pages/sec counter; then add the Avg. Disk/sec/Transfer counter from the Logical Disk object (note that this last object won't be visible unless diskperf counters is enabled as instructed above). The latter counter should point to the drive where your swap file resides.
The first counter measures the number of pages of application code that have to be read from the disk because they're not in memory, and the second one gives the average time in seconds per disk transfer. The product of these two counters indicates the amount of disk access time used by paging out to disk. If it's consistently over 10 per cent, you are short on memory.
The Server object has a number of counters that are useful to determine if you're short on memory. The Pool Nonpaged Failures counter shows how many times a page allocation from memory that's not supposed to be paged out to disk has failed; if this counter goes up, you need more memory. Similarly, a large Paged Pool Failures counter indicates that either your physical RAM or swap file is being overcommitted and needs to be increased.
The Pool Nonpaged Peak counter shows the maximum size of the non-swappable pool of memory (in bytes). This gives you an indication of how much memory you really need for your system.
PROCESSOR The Processor, System and Processes objects supply the counters that will reveal if you're short on CPU muscle.
If the Processor / %Processor Time values are over 80 per cent and coincide with low disk and network counters, you should consider a beefier CPU. This counter is a measure of CPU activity, as is the System / Processor Queue Length (values over 2 indicate congestion, with threads waiting for CPU time) and the Processes / %Processor Time counter (use this to check if one or more applications are hogging the CPU).
DISK The PhysicalDisk / Disk Queue Length counters show the number of outstanding requests for the disk subsystem. If the queue length is greater than 2 and the Physical Disk / % Disk Time counter is consistently high, your disk subsystem is creating a performance bottleneck.
Other counters worth keeping an eye on include Physical Disk / Average Disk sec/Transfer (values greater than 0.3 seconds indicate that the disk controller could be continuously retrying the disk access because of failures) and the Physical Disk / Disk Bytes/sec one (lower than 20KBps usually means you have an application that accesses the hard disk inefficiently.)