Roaming Thoughts

There is no perfect wireless network. In fact, in terms of efficiency, there aren’t any that can be said to be even good. We blast vast amounts of signal in all directions at the speed of light in the desperate hope that an infinitesimally small amount of that power will alter the voltage on the antenna of a single device moving blindly through our range. I’ve learned two things in the years since I became fascinated with WiFi. First, is to appreciate that every frame that successfully reaches its destination is a miracle of science and engineering. Second, most wired network engineers I’ve met don’t know anything about how it works.

Now let me be clear, this post is not a takedown of wired network engineers. For how ubiquitous WiFi has become in our modern lives, wireless engineering remains a very niche specialty. Many genius network architects I’ve met have gone their entire careers without so much as a glance at the technical aspects of wireless. (And to be fair, it wouldn’t take a genius to run circles around me in switch/route) But, as I see wired network engineers wading into wireless, I want to discuss what I see is a fundamental flaw in their approach. A mentality that serves them well on the wired side, sets them up for failure in wireless.

The WiFi Cliff is an analogy I developed when talking with wired engineers at a major retailer for which I was consulting. Each time an issue was raised they’d research the problem, only to be met with conflicting and contradictory advice. They adjust and then wait to see if the issue continues. They look at the problem as binary. The device is functional, or it is not. This mentality is not a bad one to have. Most of the day-to-day issues in wired networking are binary. Bad config, bad cable, bad port. You replace the offending piece and functionality resumes. There is a correct answer. There are plenty of binary problems on the wireless side as well. Sometimes the AP does burn up. Sometimes an update breaks connectivity. These problems are binary. But, where they represent the bulk of wired issues, binary problems are a much smaller part of the job of a wireless engineer.

So, What is the “WiFi Cliff?”

To help these engineers understand how their tried-and-true methodology was hurting their wireless network, I gave an example of a cliff. The user’s device is like a person standing on a cliff. As long as they are walking around the top of the cliff, rescue teams aren’t notified. The assumption is that there is no problem. Once the person goes over the cliff, then people call for rescue. A ticket is put in, the “WiFi is bad.” Rescue teams are dispatched, and the person is brought back up. The problem occurs when the rescuers look at the problem as a binary. The person is brought back up to the top of the cliff and the rescuers leave. They leave the person on the edge of the same cliff and wait for them to fall again. Once the device is functional, their work is over. No matter how close to the cliff edge they still are.

Now let’s talk about what should be done. Get everyone off the cliff. Close it down, wall it off. As a wireless engineer once told me, the only perfect wireless network is one with no clients on it. But that really wouldn’t get anything done. So, we are stuck with people on the cliff. We are left juggling imperfect options. This is why the advice those wired network engineers kept finding was so contradictory. When you’re keeping people from slipping off a cliff, the specific strategies that will work become very situational. There is no “correct” answer. But there is a different methodology that we can use to make this work. We move the people away from the cliff edge. We move them away from catastrophe. We put up sensors that will notify us if people are getting too close, so we can intervene before tragedy. We don’t stop making the cliff safer just because no one is currently over the edge. Instead we incrementally improve the network beyond the point of just functional. We build in resiliency and fine tune our networks because failures will happen. But a slip and fall isn’t catastrophic if it’s not at the cliff edge.