Next-Generation WLANs: Version 1.0 update from November 2003

It's been 10 years since thousands of IT organizations, frustrated by the limitations of desktop computers and mainframes, concluded that the path to progress lay in client/server computing systems. It seemed logical that distribution of processing across a high-speed network would let you build more scalable applications and save you money to boot.

I worked in one of those IT organizations, and it didn't take us long to realize that this transition would not be smooth--or inexpensive. We learned that networked computing systems are difficult to engineer, and that compromises needed to be made. It took several years for the technology to mature enough to make client-server computing viable.

PAST IS PRESENT

There are lessons here for IT organizations considering the emerging wireless LAN infrastructure market. Today's dominant technology--the smart access point--is akin to the PC of the early '90s. It's become increasingly powerful, and increasingly difficult to manage. There must be a better way.At least some venture capitalists seem to think so. Perhaps they see Cisco's wireless LAN model as flawed and its position as the enterprise WLAN king not quite as secure as Microsoft's dominance in the PC arena proved to be. Why else would we see upward of $100 million flowing into start-up wireless LAN companies in this post-bubble era of cautious tech speculation?

At least half a dozen start-ups--including Airespace, Aruba, Chantry and Trapeze--are rolling out enterprise-oriented wireless LAN infrastructure systems this year. Some are well-funded, well-managed and bursting with bright engineers trying to build a better mousetrap.

But it's a good bet that two years from now, most of these companies won't be around. Despite this market's significant growth potential, there isn't enough business to sustain them.

NOT AS EASY AS ETHERNETSome analysts refer to the new WLAN systems as wireless switches, but that's a misleading term that conjures up too close a comparison to Ethernet switches. Wireless LANs are significantly more complex than Ethernet LANs, at both the physical and data link layers. So while Ethernet switch designers don't have to think twice about moving bits reliably down copper and fiber cables, WLAN infrastructure designers must devote an extraordinary amount of attention to the matter.Much of the industry is focused on the differences among wireless infrastructure designs. But there's nothing inherently better about thin wireless access points, which provide a limited set of functions, than about smart access points from Cisco, Enterasys and Proxim. For a short time, some suggested that thin access points could be made--and sold--for less than smart access points, but in today's commodity chip market, that's just not the case.

The key difference isn't in the access-point design but in the way the access points are coordinated to deliver an integrated set of wireless connectivity, mobility and security services across a multibuilding network infrastructure. This coordination isn't easy to achieve--a fact that's just dawning on both the start-ups and the established players. Not surprisingly, the ambitious product-delivery schedules established by some new system vendors have slipped.

If you plan to deploy a wireless LAN in the coming year, you can invest in smart access point technology that's mature but limited or in next-generation technology that's not yet perfected. If you opt for a newer, "state of the art" system, just be sure the vendor is willing to install a pilot system at your site so you can verify both its capabilities and its shortcomings. Buying blind is a mistake you don't want to make.