Home > WLAN planning > Corner Cases

Corner Cases

February 26th, 2014

Most Wi-Fi manufacturer’s marketing departments would have you believe that 99% of all deployments are what I’d call “corner cases.” I call B.S. (as usual).

Here are the high-density/high-throughput (HDHT) corner cases that so many manufacturers would have you believe are so prevalent:

  • Large K-12 and University libraries, cafeterias, lecture halls, and auditoriums
  • Stadium or gymnasium bowls
  • Large entertainment venues (e.g. music and theater halls, night clubs)
  • Trade shows
  • Urban hotspots
  • Airports


Combined, these use cases comprise less than 1% of all Wi-Fi installations.  In other words, the opposite of what many marketing departments would have you believe. Let’s look at this from another angle. Here’s a list of use cases that do NOT fall into the category of HDHT, but may have other technical challenges or requirements, yet these same marketing departments want customers to believe they are HDHT environments.

  • K-12 classrooms*
  • Malls
  • Majority of airports

* Note: Some folks believe that one AP per classroom (or even one AP per two classrooms) is a bad idea due to adjacent channel interference (ACI) or co-channel interference (CCI), but that’s a design matter based on a long list of design criteria that can include wall construction materials, AP output power, client device type, client device output power, and MUCH more. I assert that one AP per one (or two) classrooms is a good network design in many K-12 environments, and this usually means less than 35 devices per classroom, worst case. 35-70 devices per AP (2 radios) does not constitute high-density, but may necessitate good L1, L2 QoS, and L7 handling.

Consider all of the common deployments that constitute the majority of WLAN environments:

  • Office environments
  • Warehouses
  • Manufacturing
  • Hospitals
  • Distributed healthcare facilities
  • Cafes
  • Bookstores
  • Hotels

So if HDHT handling isn’t a big deal in 99% of the use cases, what is important? If you ask that question to those same vendor’s marketing departments, they would say Performance! Once again, I call B.S.

After speaking with a variety of network administrators and managers, I’ve found it very difficult to find anyone who can produce statistics showing an AP sustaining more than 10Mbps over the course of an 8-hour business day. Even the peak throughput on the busiest APs aren’t all that high (a couple of hundred Mbps sustained only for a couple of minutes while large files are being transferred). It’s been my experience that busy branch offices, with a single AP serving 50-60 people, is where you find the most sustained WLAN traffic over a single AP.

If 10Mbps is considered “a very busy AP”, and decent 2×2:2 802.11n APs can sustain 200+Mbps of throughput across two radios given the right RF and client environment, then why is everyone talking about performance? I hear vendors bragging about their 3×3:3 11ac APs being capable of 900+Mbps of throughput under optimal conditions. While that kind of throughput is sexy to IT geeks who think that “too much is never enough”, most customers just want it to work. At 200-400 Mbps of throughput for 802.11n APs, why do we care so much about buying premium-priced 11ac APs again?

What do we get out of those 11ac APs anyway? 256QAM is useful only at short range and only for 11ac clients. TxBF is only good at mid-range, and only for thoses client that support it, which is basically none. Rate-over-range is better for uplink transmissions, but if you’re designing for capacity, voice, or RTLS, then this is of no consequence. There may be slightly fewer retransmissions due to better radio quality, but that’s mostly “who cares” also. Bottom line: don’t upgrade your 11n infrastructure for the purpose of speed. If speed (e.g. rate-over-range and raw throughput) is your goal, spend your budget on refreshing your 11ac clients first.

Customers who rush out to buy the latest, greatest, fastest AP end up paying a big price premium for a performance gain that they’ll never, ever, ever, ever use. It’s just silly. They get duped by the marketing message that HDHT handling and ultra high-performance matter in 99% of use cases, when in fact it matters in <1% of the real world use cases. Wi-Fi infrastructure technology is progressing quickly, and the PHY/MAC layers are so far ahead of typical use cases that customers should be focused on correct Layer-2 design and receiving value above Layer-2:

  • Robust, global, cloud management and services option
  • Strong security, compliance and reporting
  • Device tracking / location services
  • Social media integration (i.e. Facebook, LinkedIn, Twitter)
  • Guest and retail analytics
  • Managed services enablement

If you’re going to buy (or upgrade to) an 11ac infrastructure, there’s a very important reason to do it that is unrelated to the speed at which you move frames across the air: intelligence. Some APs don’t have the horsepower to do any significant local processing, and that leaves three options related to infrastructure intelligence:

1) don’t have any
2) send everything to the cloud
3) send everything to a controller

I prefer that APs have enough oomph to get the job done if that’s the optimal place to do the work. There are times when using the cloud makes sense (distributed, analytics), there are times when using the AP makes sense (application visibility/control), and there are times when using a controller makes sense (2008-2009). #CouldntResist

I’ll summarize all of this by asking that prospective customers of Wi-Fi infrastructure remember that they will likely never use even a small fraction of the throughput capabilities of an AP. What will have a significant impact is Wi-Fi system cost, Wi-Fi system architecture, and network design. Don’t get duped by the loud, obnoxious marketing hype around the speed/throughput. Think twice, buy once.

Don’t miss future posts! Subscribe by email.

WLAN planning

Comments

  1. No comments yet.
  1. No trackbacks yet.

Your email address will not be published. Required fields are marked *