Selina Lo, chief executive of Ruckus Wireless, has been with plenty of networking companies (NET, Centillion, Bay, Alteon, Nortel to name a few) and has an opinion on almost any topic in the industry. Over coffee in London, she told me several - but it was her 802.11n fast Wi-Fi opinions that most interested me.
She weighed in on WiMax first, though. In the wide area, it's "the revenge of the mobile carriers on 3G" and, where it overlaps with Wi-Fi, "it's like like Ethernet versus token ring over again - eventually democracy wins!"
There are places in the world that are going for WiMax in a big way, and Ruckus is experimenting with it, along with other things, she said. But WiMax was small talk. Wi-Fi is what counts for Ruckus.
Ruckus makes Wi-Fi systems, that deliver fast data. It delivered its first system in 2005 using BeamFlex, a technology close to MIMO (multiple input multiple output), and before that made radios for NetGear access points, well before draft 802.11n products were considered. It uses multiple antennas (not multiple radios) to set up better channels between client and access point, allowing better range and reliable connections.
After producing home gateways for service providers, the company added a ZoneFlex, a Wi-Fi switch aimed initially at smaller enterprises.
The Wi-Fi switch was a surprise, but it doesn't replace Ruckus' service provider focus: "We have half a dozen solid customers and another half a dozen waiting," says Lo. "For two years we have been the only Wi-Fi vendor that can do true real-time video distribution."
The Wi-Fi switch market is crowded, but there's room for Ruckus because there is not much innovation at the level where the company can add value, she says: "The industry has bifurcated - there is innovation in the chips, and there is innovation in the management of the systems. In the middle layer, it's all "cookies cut in Taiwan," she says.
The top level RF management systems have to work hard to manage their commodity access points, she says, a wasteful arrangement that means a hotel she visited recently had 81 access points to serve Wi-Fi in 150 rooms. With a controller that could cost $100,000, she says.
Hotels, you won't be surprised to learn, are a major market where Ruckus will be selling ZoneFlex, which is intended to be easy to install (no site survey) and cheap. It's moved on from the initial SME focus, she says, though it doesn’t provide all the IDS/IPS bells and whistles that Cisco or Aruba add for larger customers (they can be got with a partner, she points out). Many of Ruckus' target customers are moving upwards from single access points to a managed Wi-Fi network for the first time, she says.
"Hotels need good RF coverage and don't have a VPN requirement," says Lo. Hotel guests will normally set up their own VPNs back to corporate headquarters. She's planning to get ZoneFlex set up with a couple of distributors in each country, she says.
But what about 802.11n?
As a switch vendor, though, Ruckus needs a position on 802.11n, currently made essential by Cisco's endorsement. The trouble for Ruckus is that, with its BeamFlex, it has a lot of what 802.11n offers, but it can't afford to stand outside the hype. Ruckus is noticeably sceptical about 802.11n in its blog.
The euphoria around 802.11n is misplaced, says Lo. It will have to be more robust and have better interference handling before it takes off, she says. "Fundamentally, an 802.11n system is not one radio but three. There will be more interference. The chance of instalbility is higher." There are also too many options in the 802.11n standard, so users won't be sure what they are getting, she warned.
Nevertheless, Ruckus is doing 802.11n, but its access point, due next quarter, will be a single band AP, only in the 2.4GHz spectrum. "N is goodness, but we need it to be completely ready," says Lo. Ruckus plans to keep costs down, and single band will be cheaper for a while yet. And dual band Wi-Fi won't give much more benefit, she says.
Despite the wealth of consumer N access points available, Lo disagrees with the vendors who say N will be a consumer technology first. "Consumer devices are not picking it up," she says, largely because of the difficulty of delivering it at a low enough cost.
N is going to be more use for jobs like providing backhaul in a mesh, she suggests - and this is something Ruckus can do with its single-band device, because it forms separate beams. "We're not omni directional" she says. "Single band is as good as you need."
Against this background, vendor promises about 802.11n are "posturing," she says. Users are keen to adopt N, and vendors are keen to convince them they can provide it, "but the tools aren't there."
It's taken chip vendors longer than expected to get 802.11n products out, and there are other problems such as the lack of N-aware survey tools.
Even if these arguments hold water, won't the single band N stance look lame next year, when the price of dual-band comes down, thanks to single-chip implementations, I ask?
"We'll see," she says.