Making Sense of Evolving WLAN Standards, Part 2: Radios and Centralized Architectures
January 22, 2004
The arguments and possible 'standards' for thin access points are plentiful, as are the reasons against them. Which makes the most sense in an enterprise?
Despite the fact that the biggest obstacle to WLAN deployments in the enterprise -- security -- is actively being addressed by new protocols like WPA, enterprises are still wary of Wi-Fi.
According to IDC, a market research firm, the consumer market continues to be the driving force behind Wi-Fi. IDC argues that although Wi-Fi is catching on with highly mobile road warriors, general business use of Wi-Fi, especially within the confines of the physical enterprise, is lagging well behind home use. However, IDC is optimistic, noting that as mobile workers become increasingly dependent on electronic forms of communication, user demand and potential increases in productivity will drive businesses to implement wireless networks.
ForceNine Consulting seconds this notion, saying "purchases of WLAN hardware from large U.S. private and governmental enterprises will exceed purchases from households and small-office/home-office entities (SOHOs) for the first time in 2005." ForceNine projects that North American enterprise WLAN purchases will reach one billion dollars by 2005, with annual WLAN spending per enterprise employee reaching nearly $13 by 2005.
But ForceNine also sees excessive caution when it comes to enterprise commitment to wireless. "In a recent survey we conducted of more than 50 enterprise CIOs, we found that most large enterprises have either not yet deployed WLANs or have committed only to small pilot deployments," said Dr. Sam Book, a partner at ForceNine. "Once enterprises begin to better understand wireless security, they will be more likely to embrace real enterprise-scale roll outs. However, when security fears are assuaged, other issues will emerge, such as evolving radio standards and equipment interoperability."
Choosing the Right Radio Standard
For those intrepid enterprises willing to take the wireless plunge today, the first decision they must make is a relatively simple one: determining what flavor of 802.11 they want to deploy.
802.11b clearly has the early momentum, offering throughput of 11 Mbps in the 2.4 GHz band over a 100-foot-radius footprint. For the reasonable future, 802.11b should be sufficient for most of the tasks mobile employees need to do, such as checking e-mail and accessing presentations, documents, and the Web. When cost is added to the equation, 802.11b equipment, which is quickly becoming commoditized in regards to APs and nearly ubiquitous in laptops, looks even more appealing.
Looking ahead a few years, as deployments evolve from the pilot to the enterprise level, 802.11b's throughput may not be enough, especially if a large user base begins seeking untethered access to next-generation applications like voice over WLAN and mobile videoconferencing. For cutting-edge enterprises, 802.11a makes more sense. It offers throughput of up to 54 Mbps in the less crowded 5 GHz band, while providing twelve non-overlapping channels vs. only three for 802.11b.
In November, the FCC sweetened the 802.11a pot by releasing an additional 255 MHz of spectrum in the 5.470-5.725 GHz band for unlicensed wireless devices. This increases the spectrum available to 802.11a devices by nearly 80%. Part of the reason the FCC freed up the spectrum was to harmonize the spectrum available for unlicensed devices throughout the world.
"The bottom line, from a customer standpoint, is that with 802.11a users gain more channels, and their equipment will still work as they travel to different countries," said Paul DeBeasi, vice president of marketing at Legra Systems, a provider of WLAN switches.
While several companies, most notably Intel
, have been talking up the
many advantages of 802.11a -- despite the fact that Intel has yet to deliver
802.11a chips, but that's a separate story -- yet another 802.11 variant hit
the market with a bang last year: 802.11g. Offering the same 54 Mbps of throughput,
the main advantage of 802.11g over 802.11a is that it operates in the 2.4 GHz
band and was designed to be backwards compatible with 802.11b.
I asked DeBeasi which flavor of 802.11 he recommends. "That depends," he said. "Even though 90% of the installed base is 802.11b, customers prefer a network infrastructure that supports all three types of 802.11." A multi-band solution (supporting 802.11b/g and 802.11a) provides future-proof protection by being more flexible than an 802.11b/g-only or 802.11a-only solution. "A multi-band solution allows customers to design a network that supports all three standards simultaneously," he said.Considering that 802.11a and 802.11g offer the same amount of throughput, with 802.11g also supporting legacy 802.11b clients, why then would you still opt for a multi-band solution? Isn't 802.11g good enough? Not necessarily. The main problem with 802.11g is that it still occupies that crowded 2.4 GHz band, which must share space with such devices as microwaves and cordless phones.
"If you have any concerns about performance, or if you're experiencing a lot of interference with your 802.11b pilot project, then you'll be best served with 802.11a," DeBeasi said. In its relatively uncrowded band, 802.11a has more channels than 802.11b, so it boils down to a simple choice," DeBeasi added. "If you simply want wireless coverage as fast and as cost effectively as possible and if you want to be able to communicate with the broadest set of available client devices, then 802.11g is your best bet. However, if you want a high-performance network, you need to factor in support for 802.11a," he said.
One drawback of choosing a multi-band infrastructure is cost. A multi-band infrastructure will likely add as much as 20-30% to the initial cost of the network. What an enterprise must consider, however, is if this initial price premium will actually translate into long-term savings.
"We believe that any enterprise investing in WLAN technology would be wise to invest in a multimode infrastructure," said Colin Macnab, VP of Marketing at Atheros Communications, a provider of multimode WLAN semiconductors. "You want to be able to serve every available client, and, moreover, you need to be cognizant of the fact that unlicensed wireless spectrum is an extremely scarce resource. Even if you prefer 802.11a, you don't want to just abandon the channels available in the 2.4 GHz band. As user bases grow, each available channel will be extremely valuable."
The Difficulty of Managing Access Points
The shift from consumer-grade to enterprise-grade wireless networking is being driven by the new class of Wi-Fi switching devices. WLAN switches make it possible to manage a large and distributed base of APs, but as these switches become more common, a new problem emerges: how does your switch communicate out to your many APs? Without a standard in place that facilitates this communication, one of the main benefits of switch-based WLANs -- the ability to build multi-vendor, heterogeneous networks -- is lost.
The leading candidate for standardizing switch-to-AP communications is the lightweight access point protocol (LWAPP), which is currently going through the process of becoming an Internet Engineering Task Force (IETF) ratified standard. LWAPP centralizes many of the functions that now reside on the APs themselves. "LWAPP goes a long way towards making APs, which until recently were consumer-class products, suitable for enterprise deployments," said Alan Cohen, vice president of marketing at Airespace.
With LWAPP in place, enterprises can secure and manage large WLAN deployments as a single network, rather than attempting to manage a slew of individual APs. LWAPP would allow WLAN system vendors to focus less on switch-to-AP access software, enabling additional innovation in the higher-layer features, such as authentication, packet filtering, and policy enforcement for a switch or router. Moreover, those other higher-layer functions -- including QoS, rouge detection, and load balancing -- can be centralized for an entire enterprise WLAN. A network manager can simply install a WLAN switch in a wiring closet or in a data center, while relying on LWAPP for AP device discovery. It doesn't even matter if the APs are from multiple vendors. LWAPP standardizes the communications between all of them. With LWAPP running, the switch will automatically discover and integrate any AP into the network, enabling large-scale, multi-vendor, heterogeneous networks. "The major advantage of this," Cohen said, "is simply choice. Customers always prefer having choices."
LWAPP, however, has its detractors. Cisco
supported LWAPP before doing an about face, and now Cisco is seen as one of
LWAPP's principal opponents. According to Ron Seide, product line manager in
Cisco's wireless networking business, however, this perception is exaggerated.
"We support some of the ideas behind LWAPP, but we're more concerned with
other wireless LANs issues, most specifically the interoperability from the
AP out to client devices."
Seide said that it's too early for Cisco to commit to any specific switch-to-AP standard at this time. Cisco recently announced its own proprietary version of LWAPP called the Wireless LAN Context Control Protocol (WLCCP). Cisco, however, does not have plans to standardize this protocol.
Other vendors complain that Cisco too often pushes its own proprietary protocols. In addition to WLCCP, Cisco has been advocating the Cisco Compatible Extensions (CCX) program rather than putting any effort behind standards. Moreover, Critics argue that the company intends to get chip and NIC vendors to add proprietary features like LEAP security into their products, which, if successful, could undermine the push for broader open standards.
Seide said that while Cisco is not opposed to standards, he's not convinced that APs should be "thinned" down, as LWAPP implies. "I would argue that some functions are best located on the access points, such as port blocking, and while I agree that you'll see a thinning down of wireless domain services on the AP side, you may see a fattening of other features like QoS," he said.
Seide argued that "fat" vs. "thin" isn't really the issue when it comes to the future of APs. "It's not a problem of size, but flexibility. Different settings will dictate different deployment strategies." He used the example of an insurance agency. At the main office, a thin-AP strategy would make sense, but out at a remote claims office, for instance, where there may be less network support, more features would need to be resident in the APs themselves.