Tag Archives: measurements

Want a successful WLAN deployment? Measure twice deploy once.

I’ve recently come to the sad realization that most technical experts are using the wrong measurements to determine the success of a WLAN deployment. Don’t get me wrong; measuring things like throughput, SNR, retry rate, and authentication/re-authentication times is very important. What I’m saying is: these are all measurements used to determine if the hardware and software components are playing nicely together. Am I the only one who finds it disturbing that we claim to design networks for people to use but we don’t have a good set of measurements to determine if the ‘people component’ plays well with the hardware and software components?

It would seem to me that the success, or potential for success, of a WLAN deployment is largely determined by the end-users. If the people components don’t jive with the infrastructure components, then your WLAN deployment will fail. After all, what is the use of having a fancy WLAN if nobody uses it? Let’s look at a few of the current measurements available to us:

  1. Throughput – Using an application like iPerf, we can get some very accurate figures on just how much data we can cram through our wireless pipe. Higher Mbps values mean faster uploads and download speeds.

  2. Retry Percentages – This measurement is important because a high percentage of retry frames means something is not right. There could be a major source of interference, hidden nodes, LAN-side cabling issues or any number of problems forcing your wireless clients to have to constantly have to repeat themselves. The goal is to have a very low retry percentage.

  3. Latency – Just how long does it take your data frames to get from point A to point B? If latency is too high you’ll notice some applications start to act a bit flakey. A good example of this is VoIP: high latency leads to jitter and dropped calls. When it comes to latency you don’t want to shoot for the stars because they are far too high. Instead, shoot for the floor since it is nice and low.

I could go on, but giving a summary of all performance measurements is not really the purpose of this post. All I am trying to show is that the current measurements are only designed to give us very technical details about the infrastructure and not about the end-users. If we truly want to determine how successful our WLAN deployment is going to be, I propose a few more measurements:

  1. User Awareness Level – Are the users even aware that your WLAN exists? Are they aware of where it exists? Who to contact if they have issues or questions? Add 1 point each time you answer no and aim for a score of zero.

  2. Barriers to Entry – How difficult is it to get started? Which credentials are required and how do users go about obtaining and configuring them? Does the WLAN work with native wireless supplicants or will users need to install additional software? If your sign-up process requires a lot of technical knowledge or technical staff intervention, then you’ve got a very high score. Hint: High scores are evil.

  3. Number of User Interactions – From the moment the user decides to connect, to the moment they open their first website, how many user inputs were required? Lowering the number of items that users have to click or enter each time they connect will dramatically improve user satisfaction and adoption.

While infrastructure-focused measurments are very important, we should not allow ourselves to believe that they provide any real insights into the success of a WLAN initiative. The majority of WLANs being deployed are supposed to be making it easier for people to live a mobile lifestyle. Taking some time to think of the users, before and after the technical work begins, should be a mandatory step in any WLAN initiative. Applications and infrastructure care about Mbps, retries, EAP types, and other technical mumbo-jumbo. Users care about the experience. Get it right by measuring both and I predict a successful WLAN deployment in your future.

Dan C.

Do you have any thoughts on what should be measured regarding a WLANs performance, effectiveness, and success? I really like to hear what you have to say on this topic. Leave a note in the comments section or share this post with your colleagues if you feel this is worth further discussion.