802.11ac: More Throughput For One or For All?
In today’s highly mobile society we try to stay as connected as we possibly can, whether that be for instant messaging, email, or keeping up with our favorite TV shows and movies. This obsession for connectivity has stressed the wireless infrastructures that are installed by many organizations. Because of this many organizations are looking to update their systems that support 802.11n technologies or even the older 802.11a/b/g standards. As a consultant it is one of my jobs to help a customer understand the new technology inside and out and make sure their networks are deployed accordingly. One of the major topics discussed when going over 802.11ac is throughput. 802.11ac brings with it a substantial jump in throughput, but there is a price to pay in order to achieve those higher data rates.
Let’s go over a basic first, in wireless technology we use a frequency, or channel, to send our signal from the client to the station (access point). Think of this channel as a lane on a highway or freeway. A single lane carries so many cars per hour. Now if we add another lane in the same direction we can carry double the amount of traffic. This is repeated for each lane that you add. 802.11n and 802.11ac both allow us to combine multiple lanes to act as a single wide lane, allowing for larger traffic to pass. These channels are reflected as 20MHz, 40MHz (2 lanes – 802.11n and 802.11ac), and 80MHz (4 lanes – 802.11ac).
Ok, so we get the hole channel concept now right? So what’s the big deal then if we start combining these channels? Channels = capacity in the wireless world. A channel only has so much bandwidth to provide, you can’t create more, it’s a very finite resource. When our goal is to support hundreds of client devices in a large university auditorium for example, we want more channels as this gives us more overall capacity. If I were to deploy a true 802.11ac network for a university in an auditorium that requires say for example 6 access points I will end up with channel reuse when I avoid DFS channels in 5GHz. The channel reuse will degrade the performance of those access points using the same channel. Now if I deploy the same 6 access points with 40MHz channels I no longer have to worry about channel reuse in that auditorium.
All of this brings me to my point: more throughput for one or for all? Are we providing wireless for the greater good or for one device? Do we need to make the wireless network perform like the autobahn where if you have a nicely tuned German automobile you can drive fast or should we make it like an American highway system with a hope for a consistent goal of the same speed for all? This is the question that you need to ask yourself when deploying a higher density network today. There are times where even a 20MHz channel width (stadiums for example) still makes more sense than 40 or 80MHz channels. The fact that I can have more channels equating to more capacity which equates to better performance for all is better than having a few high speed devices benefiting from the design.
A good overview of the Cisco AP2700 for example can be found over at http://NSAShow.com/AP2700. Blake Krone and Samuel Clements stressed the AP2700 with a mixture of 100 different client devices to see how they performed. While we proved that the AP2700 could handle 100 devices (probably more but we ran out!) it doesn’t mean that you should design to support 100 devices per AP and call it good. You need to focus on the channel capacity and channel reuse to make sure it’ll meet the requirements.