One question that I get asked a lot is how to evaluate the performance of wireless LANs.
WLANs are like any other product in most dimensions. Each product has a certain range of performance, and I mean that quite literally. The key parameter we evaluate when running a benchmark isn't throughput or rate alone, but rather the throughput obtained when testing a given combination of client and access point at a given distance. We thus call this "rate-vs.-range" testing.
But we don't stop there; we're also interested in two other dimensions, interoperability and time-bounded traffic. So, in one sense, what we're doing here isn't that different from what's involved in testing and evaluating wired products. On the other hand, the highly variable nature of radio makes benchmarking WLANs one of the most challenging technical exercises likely to face enterprise IT staffers.
You'll undoubtedly want to run a few benchmarks as part of your product-selection process - where you can also test-drive the management console - and to determine how well an initial installation is performing. These exercises are useful in both debugging a given configuration and in determining where additional infrastructure needs to be installed. This is the real value of rate-vs.-range testing - since radio waves fade or otherwise become less effective as the distance between the endpoints increases, a little benchmark testing can go a long way toward providing the most cost-effective and functional installation in a given environment.
Radio waves are variable
But that fundamental variability inherent in radio makes the benchmarking of WLANs very different from the evaluation of their wired counterparts. The objective of any benchmarking exercise needs to be the elimination of as many variables as possible so that only the features and functionality desired for comparison are being compared. In other words, the only variables should be those items under test; everything else needs to be held constant.
OK, but what about the fundamental uncertainty inherent in the propagation of radio waves? It's impossible to predict, mathematically or otherwise, whether a given signal will reach its destination, and it's impossible to determine the exact path the signal might take. That's variability.
We've published a Farpoint Group Technical Note on a recommended practice for benchmarking WLANs (download PDF). This document describes the practice Farpoint Group uses to test WLANs. The Tech Note also examines future directions in this space, which are really exciting. The next challenges are the testing of multimedia applications and enterprise-class WLANs, and I'll have Tech Notes on both of those for you.
Anyway, benchmarking WLANs is important in so many dimensions - planning, testing, evaluating performance and beyond. I hope you find the Tech Note of value, and I hope to hear about the techniques that work for you as well.
Craig J. Mathias is a principal at Farpoint Group, an advisory firm specialising in wireless networking and mobile computing. This article appeared in Computerworld