April 08, 2005, 10:43 AM — I was having lunch with a colleague recently, and he opened the conversation with this rather stunning statement: "Isn't it interesting," he said, "how the entire wireless switch business lived and died in only three years?" Whoa.
Wireless switches are indeed about three years old. The first one, the Symbol Mobius Axon Wireless System (now known as the WS5000), was introduced in the fall of 2001. This product was breakthrough, moving common functions formerly resident in access points (APs) to a centralized switch not unlike an Ethernet switch. This led to the whole idea of "thin" APs, and also spurred a significant number of new products, mostly from start-up companies.
I think what motivated my friend's statement, though, was the acquisition of Airespace, one of the leading wireless switch firms, by Cisco for $450 million. He might also have been referring to the failure of Airflow, another switch company, or the fact that the assets of Legra, another switch startup, were acquired by NextHop. Hey, wireless is like the rest of high tech - very dynamic. Events like this are to be expected, and are hardly extraordinary. The WLAN switch philosophy now dominates enterprise thinking, and the concept isn't going away. Dead? Hardly. Aruba, Trapeze, Meru - the list goes on. Lots of players; lots of growth.
And yet, there is an element of truth regarding the death of WLAN switches. It's not really death, of course, but rather evolution. And there are two important evolutionary directions for the WLAN switch. To this point, WLAN switches have mostly been configured as overlay systems - in effect running in parallel with the wired LAN. The advantage of this approach is that the interface points between the wired and wireless systems are minimal, and there are usually no changes or disruptions to the wired LAN as wireless is added. The downside is that management and security, among other items, are separate. It really is like having two different networks in one facility. This is not optimal.
There are two new architectural directions now evolving to address this issue. The first is what we call a distributed switched architecture, where a central management/control appliance manages third-party Ethernet switches, which are usually already present in the enterprise before wireless is installed. In this case, no "wireless" switches are required; rather, the management and control functions needed for wireless are resident in an appliance or server. We still don't have full integration of wired and wireless management and security functions in this case, but it does allow the use of existing wired switches as APs are added.