Although sales of wireless networking products were up 90 percent in 2002 over 2001, according to analysts Aberdeen, most of the growth came from the consumer and home office sector. Enterprise take-up of the technology was slower -- a still healthy 65 percent, says InStat/MDR -- but with no growth in revenue because of price cuts. This lack of investment growth reflects worries about security, manageability and capability. Commonly, wireless in the enterprise is seen as something that's nice to have, but with a burden of responsibility that outweighs its advantages. It's also very hard to produce a return on investment figure for most wireless installations, which does it no favours in the current climate of restricted cash
But with around half of all new laptops coming with wireless as standard -- a figure expected to approach 90 percent in the next couple of years -- and USB 802.11b down to around thirty pounds, if the enterprise doesn't supply a wireless network then users do it for themselves. That's even worse than no network. Thus, enterprises will be forced to install wireless -- and will want it to be as manageable, scaleable and secure as their existing wired systems.
Such a clear opportunity has not gone unnoticed. By the middle of the year, some fifteen companies are expected to have announced products directly addressing this slice of the market. As well as major players such as Cisco and Extreme Networks, a host of start-ups are decloaking: expect companies like Trapeze Networks, Aruba Networks and Airespace to heavily promote their options.
Most of the companies have a similar approach, at least in broad outline. The access points act as gatekeepers and switches, accepting packets off the air, checking for authorisation, applying various services to them, and then forwarding them to the wired network -- of course, relaying packets back in the other direction. In functionality they look much like network switches on the edge of the network, and the mood is to manage them in a similar way. How that management function is implemented, and what the implications are for security, depends somewhat on the division of responsibilities within the network hardware.
A fat access point has most of its network intelligence in the same box: it handles most of the protocols for roaming, encryption, management, user authentication and so on. It presents the users it serves to the wired network switch as if they were physically connected. Thin access points separate the radio section from the rest: it relies on having controlling intelligence elsewhere on the network, commonly in a wired switch.
Trapeze is promoting an idea of an amalgam of fat and thin, the fit AP. The AP handles the radio side and encryption, and the network switches handle the authentication, roaming, access limiting and other by-user functions. Trapeze goes further and suggests that the whole caboodle should ideally stay in the wiring closet. Airflow says that the logical next step is integration of the wireless controller and the wired ethernet switch.
Fat access points reduce the load on central switches, at the cost of needing to be managed. Thin access points cost less and need less management -- their basic radio functions are unlikely to need attention, and the security and other features that do need regular attention are in a switch, where they've always been. Which is better in terms of flexibility and cost-effectiveness can be a religious argument: hard figures and hard-won experience are needed, and they're not here yet.
As these new enterprise wireless solutions become available, it will become possible to compare one against the other in real world conditions: so far, most of the comparisons are of PowerPoint presentations. In particular, it will be instructive to find out how the management functions integrate with other, existing network management systems, how bandwidth holds up under load, how seamless users find the security, how easy and effective redundancy is to apply -- in short, all the questions people faced when rolling out their first and subsequent wired networks. New options, such as users roaming from place to place within and outside the corporate networks, also need addressing: most vendors have ways to link a user to their authorised -- and only their authorised -- services, no matter which subnet they turn up on, but scalability and manageability remain questions.
Nobody as yet is promising a system that can replace your wired network altogether, although with a standard like 802.11a offering a potential half-gigabit throughput -- twelve channels at 54 megabits apiece -- if aggregated the potential is there. This is years away, and needs much better radio frequency control of the wireless environment, as well as new techniques such as mesh networking and, perhaps, ultrawideband. That will be the generation after this one: for now, proper buttoned-down enterprise wireless is almost ready to play.
For a weekly round-up of the enterprise IT news, sign up for the Enterprise newsletter.
Tell us what you think in the Enterprise Mailroom.