FRAMINGHAM, Mass. — With 802.11n, enterprise Wi-Fi networks are shifting from convenient to critical. They’re becoming the preferred and primary network access for users – which means IT groups have to rethink how they deploy, secure, manage and run the wireless LAN.
Evidence from the front lines of IT shows that it’s easy to squander the three to five times improvement in Wi-Fi data rate and throughput that 11n offers, as compared to 802.11a/b/g. If that happens, WLANs that were expected to easily handle the surging number of Wi-Fi clients, absorb the sharp uptake in latency-sensitive multimedia traffic, and deliver consistent, high throughput across the enterprise fail to live up to their potential.
Rapidly evolving characteristics are forcing IT professionals to rethink their approach to WLANs.
“My new [Wi-Fi] design criterion is: people reading a newspaper on an iPhone or iPad in a restroom,” says Philippe Hanset, IT architect at University of Tennessee in Knoxville. He’s only half-joking. “People are now watching multimedia content in locations that you did not expect in the past.”
Hanset has a broad definition of multimedia. “I don’t think of it as video,” he says. “It’s really about designing a wireless network based on the new applications that people are accessing over wireless.” Those can include a video chat application like Apple FaceTime, media-rich Web applications for collaboration or enterprise-sponsored social networking, and VoIP calls.
“These have unique performance requirements, like latency and error correction,” says Andrew Borg, senior research analyst in the wireless and mobility practice at Aberdeen Group. “802.11n has to be optimized as part of a well-designed WLAN infrastructure.”
(A new industry organization, The Multimedia-Grade Wi-Fi Working Group, was recently formed to address some of these issues. Initiated by WLAN vendor Aruba Networks, the group includes users such as University of Tennessee and some companies, among them SAP AG and U.S. carrier Verizon Wireless.)
At the same time, the number of Wi-Fi users is growing fast, and the number of Wi-Fi devices per user is exploding, too. At University of Tennessee, Wi-Fi devices have grown from 3,000 to more than 9,000 in a couple of years. Today, the majority are devices like iPod touch, and smartphones instead of laptops. These new devices typically have smaller, less powerful Wi-Fi radios with less sensitive antennas than those embedded in laptops.
Dealing with these changes requires knowing what applications your users will run, today and in the future; designing a WLAN that can meet specific throughput targets to support those applications; and creating an “infrastructure” that goes beyond just the access points and controllers to include network and user security, end-to-end network management, continual monitoring, and a trained WLAN response team and help desk.
Here’s what you need to be aware of to take your enterprise WLAN to the next level.
Design for capacity not coverage
With 802.11n, a relatively few access points can create a Wi-Fi blanket wall-to-wall in the enterprise. But with the evolution of traffic types, applications and clients, that’s no longer enough. At Scripps College in Claremont, Calif., a single 11n AP could cover a six-classroom building, for example, with perhaps 25 students in each room. “But while everyone can ‘get wireless,’ I’m not sure that 150 people on a single AP, even a two-radio AP, will meet even basic performance needs,” says Jeff Sessler, director of IT at the college.
“Designing and building for capacity means taking into account the actual use-case of a given area, and deploying enough APs to meet your performance expectations,” he says.
A key element in such an approach is an understanding of how access points perform under the expected types and volumes of traffic, number of users, and applications. Thorough testing will clarify client requirements, what the AP can deliver, and how and where to deploy them to meet service level requirements.
One practice has been to deploy Wi-Fi networks that have traffic prioritization combined with bandwidth limitations for applications or users; this is part of an overall focus on constantly optimizing wireless performance. However, designing the WLAN from the outset for the emerging multimedia challenge may be a more effective long-term solution. (VeriWave Inc., a WLAN test vendor, has a white paper on best practices for pre- and post-WLAN deployment verification.)
Treat 802.11n migration differently
An 11n network is not just “faster Wi-Fi.” It comes with higher speeds, but also higher expectations. To meet those expectations, IT groups may need to formalize new deployment schemes, or update existing ones.
“We’re treating our 11n migration differently,” says David Morton, director of mobile communications strategies at University of Washington in Seattle. “It’s a different architecture and standard, and a different deployment methodology.” As part of the university’s 11n shift, the IT group updated its Wi-Fi deployment guidelines to stipulate details including which access points were being used, what signal levels were expected, and how the APs were mounted, cabled and powered.
“I’ve found that 802.11n [radios] have a significantly improved range over 11g, and consistently maintain higher performance at a given distance,” says Sessler of Scripps College. At the same time, many IT groups are using the 5 GHz Wi-Fi band for the first time on a large-scale: It has different RF signal propagation characteristics that need to be factored into an 11n WLAN.
At University of Washington, the phased 11n upgrade is overseen by a project manager. “A lot of the issues are not technological ones,” Morton says. There’s a lot of coordination to be done, and that includes getting the right people on board at the right time, he says.
Don’t forget the back-end
The best designed WLAN can be crippled because back-end services are overlooked. Two examples are RADIUS servers and DHCP servers that for various reasons start fumbling when hit with a flood of Wi-Fi requests.
In addition, DHCP servers often are not aware that a Wi-Fi user has disconnected, or may not release IP addresses in a timely manner. That can lead to the WLAN running out of IP addresses.
Manage WLAN management
Wireless management tools, compared to those for wired networks, are still immature. They tend to focus on specific problems, tend to be reactive, and lack an end-to-end perspective, according to IT staff.
“There are good Wi-Fi tools out there, some very specific to designing or troubleshooting a Wi-Fi network,” says University of Washington’s Morton. “But what are not readily available are the tools to show us what’s happening on the network, and its impact. We don’t have an overall landscape picture. And some of that is very important, as you look at making changes in the network or investing in it in the future. You want to be able to spot trends as they emerge out of the gate so you’re not caught flat-footed.”
Best-in-class companies know how their wireless networks are performing and how healthy they are, according to research by Aberdeen Group. They’re more likely to make regular, routine use of RF spectrum analyzers, for example, and far more likely to use tools for creating custom reports of network data.
Apart from having the range of necessary tools, and the expertise and expectation to use them, one key practice is to collect and use the information gleaned from these tools. Aberdeen Group has found that enterprises with the best performing WLANs are far more likely to foster knowledge-sharing through a centralized storehouse of WLAN data. It’s a key enabler for many of their WLAN best practices, according to Borg.