At Interop in Las Vegas last week, a vendor panel discussion on the "Great WLAN Architecture Debate" turned into...
an argument about cell-based topology.
Some of the vendors criticized the "channel blanket" approach adopted by Extricom and Meru Networks, forcing David Confalonieri, Extricom's vice president of marketing, to defend his company's architecture. Meru was not represented on the panel.
Many WLAN vendors advocate cell planning, whereby access points use different Wi-Fi radio channels in order to avoid co-channel interference with adjacent access points. Since each access point is using a different channel, the handoff of clients from one to another is complicated. Vendors often differentiate themselves by demonstrating how smoothly they can achieve that handoff.
Extricom's access points use the same channels. Each access point runs on several channels in order to create several blankets of seamless Wi-Fi coverage. A controller switch manages all the packet switching and associates directly with the clients. Extricom's access points don't have MAC or IP addresses. This simplifies handoffs between access points and allows customers to assign users to specific channels to help guarantee service levels.
Keerti Melkote, founder and vice president of marketing for Aruba Networks, said having multiple access points running on the same channel risks corrupting the signal, making it unusable. He added that the IEEE and the FCC are constantly adding new channels and bands into the 802.11 spectrum specifically to increase capacity. He said a single-channel access point "just will not scale."
"You don't need to crowd one channel with all your clients," Melkote said. "You can spread them across the band and give them more capacity and not violate any standard in the process"
He said that Extricom's single-channel approach makes multiple access points appear as a single access point to a client. "By making it look like one AP [access point], you are violating many, many rules."
Confalonieri took exception to Melkote's "direct stab" at his company. He said it was "categorically false" that Extricom uses a single-channel architecture.
"We are not a single-channel architecture. We are a multi-channel architecture," he said. "We are using multiple channels. We are simply saying each channel is available on every AP."
Confalonieri also rejected the claim that using a single channel for those applications wouldn't scale.
"It's easy to throw out the 's' word, 'scale,' and say single channel will not scale," he said. "But there is nothing intrinsically unscalable about using the same spectrum on adjacent sets of APs. We use the spectrum in a way that allows us to create multiple links on the same channel at the same time."
Melkote said he didn't understand how Extricom could add more devices to get more capacity.
"You have one channel that is 20 MHz wide, and there is only so much juice you can get out of it," he said. "If you want more juice, there's a lot more juice in [the rest of] the spectrum. Use it."
"Even if you can play games at the access points, you don't have control on the client side, so there's no way to be standards-based with that," said Kurt Sauter, director of product marketing with Xirrus.
Confalonieri suggested that the other vendors simply did not understand Extricom's architecture. The argument was left unresolved because the moderator, IDC Director of Enterprise Networking Research Abner Germanow, wanted to get the panel back to its original agenda. Approached after the session, Germanow said the jury is still out on whose approach is best.
"What you have are two fundamentally different architectures around the ability to have multiple cells that all work together and change dynamically, as opposed to having several layers that are a little more static but more consistent across a particular area," he said. "The question is essentially: Which ones are more scalable and which ones are easier to manage?"
Germanow said he isn't a network tester so he can't answer those questions.
"What's interesting about the WLAN space is that there is still a tremendous amount of room for innovation here, so we will continue to see companies changing their architectures and being very aggressive with the different ways they deploy wireless LANs," Germanow said. "The important thing for the customer is to sit down and think about what are the applications they are going to run [on the WLAN] and assume they are going to run more applications than they think. Assume they are going to end up running more intensive applications like locationing and voice, and bring all the vendors in to talk about what's appropriate in their environment."
Let us know what you think about the story; email: Shamus McGillicuddy, News Editor
Dig Deeper on Wireless LAN Implementation