Wireless LAN architecture experienced a revolution about six years ago when enterprises started to abandon so-called...
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
fat access points for thin access points. But now the wireless LAN pendulum is apparently swinging back the other way. Access points aren't necessarily fat again, but they are certainly getting husky.
In the move to thin access points, Vendors like Airespace (now a part of Cisco Systems) placed heavy focus on the wireless LAN controller, which managed APs that were more like basic radios. This strategy resolved many of the management headaches that networks with fat access points posed.
"The biggest [problem] was RF management," said Michael Finneran, principal of consultancy dBrn Associates. "It was way easier to install a network with [a wireless LAN controller]. In the old days, we would have to physically locate access points, manually set the transmit power and then test to be sure that we were covering the entire facility and that they didn't have co-channel interference. With the controller-based model, all that stuff happens in milliseconds and automatically. Virtually all the enterprise-grade solutions have gone that way."
But the move to 802.11n has brought fat access points, or at least slightly less thin access points, back into favor.
The high throughput speeds offered by 802.11n demand a less centralized approach to wireless LAN architecture, according to Devin Akin, chief Wi-Fi architect for Aerohive, a vendor known for its controllerless wireless LAN infrastructure. Aerohive's fat access points work cooperatively with one another to manage the control and data forwarding planes of the wireless network.
"With the high throughput of 802.11n, enterprises want to be able to count on Wi-Fi for primary connectivity, but they need high availability," Akin aid. "There is a need to make access points smart again. Reliability means no single point of failure. No bottlenecks. Traffic needs to go directly from the access point to its destination."
Fat access points are coming back, but wireless LAN controllers are also here to stay
Paul DeBeasi, vice president and research director at Gartner, said controller issues like single points of failure are overcome by installing redundant controllers. And while Aerohive claims that controllers are expensive, DeBeasi said many controller-based vendors address that by offering discounts and special product packaging.
Still, the debate on how to distribute functionality between access points and controllers is far from settled, according to Craig Mathias, principal of consultancy and testing firm Farpoint Group. The more likely answer is some combination of fatter access points and some centralized management.
WLAN controllers originally took over all foundational aspects of wireless LAN architecture: the management plane, the control plane and the data plane, Mathias said. Over time, vendors have recognized that it doesn't make sense for all three of these planes to reside on the controller.
"The data plane needs to be fully distributed," he said. "That means access points have gotten thicker because many vendors' products are capable today of directly forwarding traffic where it needs to go rather than just sending it to the controller."
But vendors have also recognized that the management plane must be centralized, Mathias said. Even Aerohive has its centralized HiveManager network management software. The debate remains open over just how centralized the control plane should be within wireless LAN architecture. Aerohive and Proxim Wireless have distributed this functionality. Most other vendors keep the bulk of this functionality in their controllers.
With spectrum analysis, Aruba access points get smarter, not necessarily fatter
Aruba Networks this week introduced a new Spectrum Analysis feature in its AP-105 and AP-120 Series 802.11n access points. By enabling individual access points to conduct spectrum analysis, the feature theoretically eliminates the need for wireless network engineers to walk around with a handheld spectrum analyzer looking for sources of interference.
"We worked very closely with our chipset vendor to make sure we could extract [spectrum] information from the chips and that we had enough processing power available to do [the analysis]," said Mike Tennefoss, head of marketing at Aruba. "Over the last two and a half years, all the 802.11n access points we have been shipping have been spectrum analysis capable. What's been missing is the software upgrade in the controller. What we're announcing now is the software module in the controller that unlocks spectrum analysis on the access points."
The software allows a network manager to turn an access point into a spectrum analyzer either temporarily or permanently, allowing for real-time troubleshooting or forensic analysis of the RF spectrum.
"There's a combination of work done by the chipset to present the data, and then the access point has to format it correctly in real time and forward it to the controller, so both [the controller and the access point] are involved in the process," Tennefoss said.
Although some might point to this new feature from Aruba as an example of a leading vendor moving toward fat access points, Mathias said the spectrum analysis feature has nothing to do with architecture.
"Spectrum analysis is an adjunct feature," he said. "When we talk about wireless LAN [architecture], its fundamental objective is to connect users to the rest of the network. Spectrum analysis doesn't do that. It's an optimization tool."
However, other features are finding their way onto the access points of Aruba and other leading vendors, whether it be spectrum analysis, firewall capabilities or policy management.
Wireless LAN controllers or fat access points: Enterprises just want something that works
In the meantime, the debate over whose architecture is best doesn't necessarily affect most network managers directly, Mathias said.
"Most users don't evaluate architecture," he said. "Instead, they run benchmarks and empirically evaluate architecture as opposed to analytically evaluating it. We're not at the point where we can actually say that one strategy [for wireless LAN architecture] is better than another. I've tested all of them, and they work."
Let us know what you think about the story; email: Shamus McGillicuddy, News Editor