802.11n: Ethernet extension or replacement?

Do you want wireless LANs to remain a mobile adjunct to your wired Ethernet? Or are you looking to 802.11n someday becoming a bona fide replacement for Ethernet in your corporate access network?

Depending on your vision, you might choose varying products, architectures, and migration paths.

Many of the WLAN vendors contend that 802.11n, with its 100Mbps+ throughput, will eventually become the “new Ethernet.” If that is the case, in addition to comparable performance and capacity, the wireless network will require availability and reliability traits on par with the wired network.

So a number of vendors are building access points (AP) containing multiple radios to back up one another, for load balancing and for integrated rogue detection monitoring. The likes of Aruba, Motorola, Ruckus Wireless and others are also building in mesh capabilities, which eliminate additional cabling and use special protocols that “route” around failed or congested APs over the air for greater reliability.

From a migration standpoint, both Aruba and Meru offer 802.11a/b/g APs that can be upgraded to 802.11n later, as user budgets allow, using a software key. Depending on purchase volume, the idea is that you basically pay today for the cost of the 11a/b/g AP and then pick up the differential between the cost of the a/b/g AP and the 11n AP down the road. There is approximately a 25% cost differential between a/b/g APs and 11n APs.

Upgrading the APs can be only part of the cost, however, depending on architecture. Controller upgrade costs, required to adequately aggregate and forward encrypted 802.11n traffic from many APs, can be significant. A controller-less architecture could eliminate those costs. Startup Aerohive, for example, notes that its dual-radio, pre-802.11n AP lists for $1,299, which is a typical price for an 11n AP. However, the Aerohive AP also contains embedded control features so it requires no centralized controller(s) and related capital costs.

Such capital costs are only part of the equation. For example, Meru’s single-channel architecture eliminates channel planning and co-channel interference, which can be worth a lot in OpEx savings. And what is your AP planning strategy? If you run all 11n APs and they “fall back” to the speeds of existing 11g clients, are you really getting your optimum 11n ROI? Perhaps consider a multi-radio architecture that segregates the 11n traffic from the 11g (and 11a) traffic - such as Extricom’s - so that all networks are running at their optimum performance.