I came across a recent blog by "Frugal Geek" Matt Ryan who wrote about the benefits of Wi-Fi and how it can help to dramatically cut costs for small businesses. While I admit that I had once called for the adoption of wireless networking to reduce infrastructure costs, my thinking on this front has mellowed somewhat over the last two years. The truth is that many of the cost advantages envisioned by tech-savvy users are probably not applicable for anything beyond very small office setups that are between 10 to 20 nodes. Beyond that, the deployment cost of rolling out Wi-Fi can go up substantially. There are a number of reasons why this happens and I've listed a couple of them below.
Wireless networks with more than 20 nodes running off a single access point (AP) bear a higher risk of getting saturated by network data. What many IT professionals aren't aware of is a Wi-Fi AP operates in half-duplex over what is essentially a single collision domain, which makes any assertion on "similar performance" against a switched, full-duplex Ethernet network a poor one. As you can imagine, activities such as video or music streaming, as well as VoIP applications will strain a wireless network far more than a wired connection. Obviously, a publishing house or architecture design firm will see network saturation points reached much earlier given the large files that get shuttled across its network. And yes, Windows workstations tend to be broadcast-happy, which can bog down an improperly designed wireless network.
The Range Limitation
We all know that the maximum advertised ranges of Wi-Fi APs are like the mythical numbers you see published in salary surveys: Someone out there has such a salary, but it's definitely not you or anyone else you know. Using a range expander to increase the reach of a wireless network is a suggestion that is often mooted, though anyone who recommends them have obviously never used one. They don't work well in my personal opinion and the only one I've owned is sitting in a box at the moment. I suspect that their poor performance is also the reason it's not easy to find one to buy these days, and no enterprise vendors ever carry them. Since most range expanders cost about the same as another AP, it actually makes more sense to just get an additional AP.
Addressing the Above Issues
The above issues are just some of the key considerations when setting up a Wi-Fi network. The logical solution would be to increase the number of APs, which does pose its own management and interference-related set of problems once you scale beyond two or three APs. Central management of APs puts you in enterprise territory, where the cost of each AP magically quadruples (or more) and generally requires a four-figure sum for the cost of a central controller appliance.
And even in an average-sized network of 25-75 nodes, it makes security sense to deploy authentication schemes such as 802.1X/PEAP or the use of an internal VPN server. Throw in the appropriate fail-over server for the above-mentioned services, advanced routing functionality to manage VLANs and routing between subnets (to address broadcast-related issues), and get back to me again on how Wi-Fi is the cheaper option versus some Ethernet cabling work and a cheap Gigabit Ethernet switch.
Now, I am not rescinding my position that Wi-Fi can bring about cost savings, and Wi-Fi does indeed afford a unique set of advantages and convenience. However, I think it would be erroneous to paint an all-too-insouciant picture of cost savings for anything but the smallest networks. As with practically any technological decision, a move to adopt Wi-Fi (or not) must depend solely on one's specific needs and not because it looks like the cheapest option on the table. For businesses running a mid-sized network, deploying an enterprise-grade Wi-Fi infrastructure that can adequately and reliably support business needs can actually cost more than a wired-only installation.