For years, society has been consumed by the desire to go faster and over time that concept made its way into the networking realm. Anyone that has been around technology for any amount of time is no stranger to the concept of “speeds & feeds” surrounding networking gear. Simple comparisons made by others to exaggerate the ability of hardware, that have profound impact on how we buy technology.
Enter wireless networking. Enter the same “speeds & feeds” sentiment. Enter marketing. Yes that right…Marketing. One of many words in the engineer’s lexicon today that can make even the most seasoned network engineer cringe. I recently took a walk through the wireless networking isle at a local big box retailer (you know the one with the yellow tag in their logo?) and was amazed at what marketing has done in the consumer space. A TRI-BAND, WHOLE HOME ACCESS POINT with “Four external antennas maximize your Wi-Fi signals for increased range in your home, and the quad-core processor delivers wireless speeds of up to 7.2Gbps.” WOW!!! WHERE CAN I GET THIS MYTHICAL DEVICE??!?!? *Sarcasm Implied* Get real. Even if the world was completely barren and you were the only RF signal out there, the realistic chances of ever obtaining that kind of throughput are <insert favorite comparison here>. LOL! No that wasn’t a typo, I simply couldn’t think of something with odds that bad 🙂
The premise of this post comes from a recent discussion between several WLAN professionals and how best to test the head to head capabilities of enterprise class wireless hardware. While I agree that in some cases, selecting the AP with the highest amount of potential throughput/processing power is critical for success, for many this just simply isn’t the case. I would argue that the capabilities of the AP hardware can only be valid in the context of the complete system, and not of the singular piece of AP hardware. After all, how many enterprise class users run a single AP in their building that faces ZERO contention from other RF sources in the environment? I suspect that answer is pretty close to 0.
With all of this being said, are WLAN drag races helpful to the professional community?
My answer is NO. While I will always withhold a firm 100% guarantee of my answer (which any engineer worth their salt should do), in many cases the info gained from one of these tests is highly subjective to the testing conditions present. The results simply aren’t something that is applicable to the real world use cases of enterprise networks.
Reasons I dislike WLAN drag races.
- A professional network should be designed so that ALL areas of the desired coverage area have both primary and secondary AP coverage. This serves to provide redundancy in case of AP failure, as well as additional spectrum for devices in that area to use in case the primary AP has become saturated. It is a physical impossibility for a singular RF channel to carry more than X amount of capacity. That’s right folks, even the great mythical unicorn single channel architecture (SCA) can only carry so much capacity in a channel before it physically reaches its maximum. (For more info on SCA adventures check out https://badger-fi.com).
- Contention-less environments DO NOT exist in the real world. Last time I checked, most buildings weren’t faraday cages, that withheld the ability of users to bring in outside items such as cell phones & Mi-Fi devices that can quickly pollute the RF environment with tons of ACI & CCI. Contention plays a HUGE role in the speeds & feeds a WLAN is able to provide to its users, and even the smallest of energy sources provide the potential to interfere with wireless networks by inhibiting devices from communicating. (For more on how this inhibits communications check out my earlier blog post on playing the game)
- No two vendor APs are alike. With the myriad of WLAN chipsets, antennas, processors, memory types etc, it is impossible to have identical AP from multiple vendors. Heck, even two or more AP from the same vendor, with the same model # etc can’t be the same due to the inability of calibrating the devices. (I do applaud Ekahau for working hard to do this with their new Sidekick device. It is definitely a step in the right direction!)
- The software makes up a huge part of the network and obviously that’s different between vendors as well. In some cases, the hardware/software combination even produces results that favor download throughput vs upload since the vast majority of communications in most networks are loaded to the client receiving side.
- Testing methodology matters greatly. While I won’t go into great detail here, I’ll simply say that there is a reason for centuries scientists have used the scientific method. We all know that one simple change in the environment throughout the course of the test can greatly skew the results in one way or the other. Averaging results provided from testing in real world scenarios is vital to being able to prove the validity of such tests. Peer review is necessary!
These are just a handful of reasons I could think of off the top of my head. While there are many more, I will close with two final points.
Vendor defaults should never be used during testing or production. Multiple testing methodologies have been used over the years for comparison testing, but no vendor has the ability to create default settings on their product that will work the same across multiple environments. There are reasons that products have so many “nerd knobs” these days and tuning each of those appropriately to the environment the network resides in is critical to network performance.
Comparison tests between multiple vendors products should be taken with a grain of salt if all parties aren’t present and allowed to configure their gear. The nerd knob comment applies here as well. Each vendor has different recommendations for different scenarios and performance settings could vary greatly between a classroom environment vs VHD LPV environment.
At the end of the day, it doesn’t matter that you have the hardware/software combination that has been shown in a test to outperform others in a drag race. What is important is that you have a properly designed WLAN, and as such the network is able to meet your customer requirements. Anything extra is just icing on the cake.
Any feedback is always welcome. I’d love to hear your thoughts!