In my last post, I talked about what you need to get started building out your wifi network. The main points were adequate switching and some basic network services like dhcp. But the most important item from the list is a dedicated network team. If you meet the first part, you probably have the second part to thank for that. Unfortunately, while you might have some great people working on your network, if they’ve been around a while this whole “wifi” thing might be a bit scary for them. Today’s post covers two areas. First, a few basic things that will help your network guy put wireless in perspective. Second is the different stages of your wifi deployment.
Basics
When we talk about wifi speeds, we tend to talk about them in the same terms as we do for wired networks. This is misleading. When we talk about a 100mbit or 1000mbit (1gbit) wired network, this usually refers to a switched network. Wireless access points are not switched.
On a switched network, each packet moves between a computer and a port on a switch, or perhaps between two switches. It’s the job of the switch to transmit a packet only when the line is clear, and only to the necessary ports. This way, it’s like each computer is on it’s own private network. You never have packet collisions and you leave as much of your network clear as possible. This means you get the most throughput possible out of your network.
On an unswitched network, all packets go to all ports and visible to all computers. If two computers want to send a packet at the same time, you have a collision. Both computers must resend, and that particular bit of bandwidth used for the broken transmission is completely wasted. This means that actual network throughput is much lower than the theoretical potential. However, networks speeds are still advertised in these terms because there’s no good way to know how much lower the real value will be. In a simple home network with few machines and little interference, you might get something close to what’s advertised. More machines makes things worse quickly, and too many machines can even push this number all the way to zero.
The good news is that your network is only unswitched within individual access points, and there is some provision in wifi to allow multiple access points in the same space. There are 11 available channels in the b/g wireless range, and more in the 5ghz range used by a and n radios. The bad news is that these channels tend bleed into each other and overlap. For example, in the b/g range you should really only use channels 1,6, and 11 most of the time (there are exceptions where you can use more). That’s only three channels. Since a single access point can generally only handle about 20 simultaneous clients, if you have more than 60 users in an area you could be in trouble.
There is one more important consideration to look at as your network grows: broadcasts. Broadcast traffic can be really disruptive to wireless networks, and so it’s important to keep your broadcast domains small — typically no more than around 500 devices, though the explosion of mobile devices (that are on for shorter periods at a time) has perhaps pushed this number higher. This is why you need equipment with good vlan support. You will need to create multiple vlans (segmented by area) as your network grows beyond this number.
At York College, we are right on the cusp of needing an additional vlan. We have well more than 500 devices. Currently we have avoided this route, because many are devices are mobile (we have an iPod program) and we have some other things in place as incentives for students to use the wired connections available in their rooms.
Stages
When I think about stages of wifi deployment, I like to think in terms of three “C”s: Coverage, Capacity, and Clarity.
For the first stage, you are working just get a basic level of coverage to an area. Imagine you are the only person on the network. If you go to a specific location, would you be able to use the network there? While building out coverage, you may find yourself tempted to look for stronger access points that can cover larger areas. I urge you to avoid this, for reasons that will soon be clear.
Once coverage is established, you begin to worry more about capicity. Sure, the access point on the floor below may be able to send a signal up a level, but if it typically already has too many users you may find you can connect but not actually get any work done. It’s time to start adding more access points beyond the basic coverage level, because you need more Capacity.
As you add access points to meet your Capacity needs, you may find that in some places you have a Clarity problems. Users at the edge of the range of an access point on a particular channel will send transmissions that reach area covered by the next access point over on the same channel, causing packet collisions and hurting throughput. Again, you may be able to connect to your network, but find that you are unable to get any work done. This problem is also called “Density”, but I find the term “Clarity” makes it easier to understand where we are in our deployment.
At this point, people tend to start worrying about things like rogue access points or interference from neighbors, cordless phones, and microwave ovens. While that’s noble, it’s mostly a mistake. The largest source of interference on your network is usually your own access points and devices. To counter this, you want to look for access points that have smaller coverage areas and lower (or adjustable) power antennas, not larger coverage areas or more powerful antennas. It may seem counter-intuitive, but this way you can have smaller size cells and you can put more access points in the same size space- you can have a more dense deployment where the frequencies you need stay clearer. The downside is that this adds expense and management issues, but it’s how they do it with the big enterprise deployments, too.
Next time, we’ll cover site surveys and access point placement.