Away from Corner Cases: High Density, Low Throughput Wi-Fi
by Devin Akin
In my blog called Corner Cases, I mentioned that high density, high throughput (HDHT) cases are in the extreme minority (<1%). In this blog, I would like to discuss High Density, Low Throughput (HDLT), which I believe will be the situation that over half of the installed Wi-Fi infrastructures of the world will face at some point over the next 5-7 years. I want to clarify that that when I use the term “high density”, I’m referring to client density (lots of clients in a physical area), not AP density (lots of APs in a physical area).
Unless you’ve been camping out under a rock, you may have heard the term “Internet of Things” or IoT for short. This moniker refers to the movement toward connecting previously-unconnected devices onto the Internet. To clarify, things are being connected to the Internet, thus we get Internet of Things. So how many of these things are we talking about? Oh… a few I suppose. Gartner is saying there will be 26 billion IoT devices and an additional 7.3 billion smartphones/tablets/PCs by 2020.
The vast majority of these devices will connect wirelessly, so we’re about to see a crazy explosion in device density. Obviously it doesn’t all grind to a stop in the year 2020, which is truly just around the corner.
The important point to make here is how device density affects: 1) network design, and 2) the type of equipment you purchase to appropriately support your customers (over the lifecycle of your next infrastructure upgrade/refresh). Most vendor marketing departments like to tightly bind high-density and high-throughput requirements together, but they are completely separate topics. You can have the following scenarios:
- Low Density, Low Throughput (LDLT)
- Low Density, High Throughput (LDHT)
- High Density, Low Throughput (HDLT)
- High Density, High Throughput (HDHT)
HDLT: the de facto standard
I don’t think that comes as a surprise to anyone. In the Corner Cases blog, I specifically addressed HDHT networks and pointed out that they are in the extreme minority today. HDLT networks are reasonably common today, but usually not to any extreme. When IoT bears its full weight on the market (which will be far sooner than you might realize), HDLT networks will be the de facto standard. In a nutshell, this means that APs will need to associate (connect) lots of devices (I foresee 100+ devices per radio becoming common fairly soon), but the traffic to/from each of those connected devices may often be sparse. APs will likely need good QoS, a good understanding of client behavior and needs, and of course security will be all-the-more important with the breadth of devices connecting to the network.
Let’s consider a specific scenario, the average branch office (perhaps real estate or insurance) with 20 employees, to make my point. Today, the branch could possibly have the following devices connected to the Wi-Fi infrastructure:
- Printers (let’s hope not, but you never know)
Let’s fast-forward to the year 2020 and consider how that same branch office might look from a technology standpoint. What items within the office could feasibly be Internet-connected in addition to what they have today?
- Security cameras
- Printers (they definitely will)
- Digital signage
- Digital picture frames at workers’ desks
- Appliances (e.g. refrigerator, water cooler, coffee maker)
- Cars that are within range of the in-building (or outdoor) Wi-Fi
- Wearable technology (watch computers, eyeglass computers, etc.)
- Building controls (thermostats, security systems, fire systems, etc.)
I’m sure I could go on and on, but for the sake of time, I’ll stop listing things. I’m sure you got my point. It’ll be a ton of things for sure. Some will want some bandwidth (e.g. picture frames sucking down 3MB photos from a file share on a server at a pace of 1 new photo every 5 minutes times 10 picture frames in the office), and some will want very little (e.g. your digital watch updating you on the temperature outside). All-in-all, the bandwidth requirements will be modest at best, but the number of devices will be ridiculous.
Remember how BYOD started? Companies tried to stop it by creating company policies. Yeah, that worked out… NOT. It will be the same way for IoT. It will progress like this:
Users: We want our things on the Wi-Fi.
Users: Yes, because if you don’t, _________.
Admin: OK, you win, but your devices will be firewalled, rate-limited, and highly controlled.
Users: I don’t care so long as they work properly. Hey wait, why doesn’t my picture frame work properly. It probably needs more bandwidth. Fix it.
Users: We’ll tell.
Admin: Ugh! OK, it’s fixed, now leave me alone.
BYOD stands for Bring Your Own Device, and trust me, they will, but not just smartphones, tablets, and laptops. They’re going to bring Internet-enabled pens, shoes, and heart monitors. You, Mr. Admin, will be powerless to stop it. You thought all of this BYOD stuff had just about fizzled and was limited to just a few vertical markets didn’t you? Ha. It’s barely even begun, and you haven’t seen complexity yet… just wait. How will you manage those Internet-enabled pens again? No, I don’t mean just at Layer-2… that’s the first step. I mean at Layer-7 also. Sorry I had to break that news to you. Bumpy ride ahead.
There are companies today who are building cloud infrastructures that are specifically designed to manage all kinds of IoT devices for the manufacturers who make them. That’s good thinking. Not every company in the world wants to build a cloud to keep their Internet-enabled devices up-to-date and to push content to them.
In closing, I will reiterate that it will soon be the number of devices, not high throughput, that will become the more significant issue across a large section of the Wi-Fi market as a whole. Make a note, it’s coming.