Benchmarking Server Performance
by Heidi Brumbaugh
This article talks about benchmarks: what tools are available, what people are saying about them, and why, if your site handles moderate traffic, you may not need them at all.
The What, Why, and Whether
If part of your job is measuring your Web server's performance-or measuring prospective
servers against each other-you may want to turn to Web server benchmarking tools to
gather the data you need. This article talks about benchmarks: what tools are available,
what people are saying about them, and why, if your site handles moderate traffic, you
may not need them at all.
What is a Benchmark?
A benchmark is simply a way to measure system performance. Benchmarks aren't new.
Engineers have used them for years. Designing benchmarks to measure chip performance,
for example, is a science unto itself. The idea behind benchmarking is simple enough.
Perform a process that is typical of what the system you're testing will be expected to
perform. Execute and time the process, then perform the exact same test on different
systems and measure your results. The devil, in this case, is in the details.
As an analogy, say Zolo the Great and Omar are competing knife throwers at the carnival.
As Webmaster, er, ringmaster, you have to decide who has the better act, so you decide to
benchmark their performance. The first step is to make sure test conditions are identical.
Are the models of the same height, in the same position, and standing perfectly still? If
they're not standing perfectly still, is the amount of wincing and squirming the same for
both throwers? Are the number and position of balloons (or other markers) constant? Is the
lighting the same (Omar might just cry foul, for example, if he had to test in the afternoon
when the sun was in his eyes.)
Once you've set up and run the tests (more than once, of course, to allow for statistical
variations) the job is far from over. You now have to interpret the results. Which elements
of the performance are the most important? Omar may claim precision should be the
determinant, while Zolo the Great argues number of knives thrown per minute is the true
measure of skill. Third parties, such as the model, may have different opinions altogether.
[Move on to the next part of the article]
Web Developer Site Feedback
Web Developer
Copyright © 1999-2001 internet.com Corp.
All rights reserved.
