Extreme Hatches Switch Surprise
The company is expected to make an official announcement of its 802.11 wireless LAN products in late April or early May, around the time of the NetWorld+Interop trade show in Las Vegas. That event is expected to provide a venue for a blizzard of 802.11 announcements from startup vendors, including Airespace Inc., Aruba Networks Inc., Trapeze Networks Inc. and Vivato Inc., among others.
Extreme’s wireless products are being developed by a "stealth department" within the company. Even other employees at Extreme have been kept in the dark about its activities, sources say.
The department is being headed by one Vipin Jain. Although not listed on Extreme’s Website or in its directory, Jain has been working in stealth mode as VP and general manager for Extreme’s "LAN Access Business" for over a year, Unstrung has learned. Prior to that, he was cofounder of Telseon Inc., an ill-fated metro Ethernet service provider (see OnFiber Takes Over Telseon).
Jain also is a well-known Institute of Electrical and Electronics Engineers Inc. (IEEE) maven and inventor of the IEEE 802.1x authentication protocol used in wireline and wireless Ethernet networks. He could not be reached for comment on this article.
Details of Extreme’s 802.11 wireless product are fuzzy, but most sources agree that it is an 802.11 "switch" – basically, an Ethernet hub or switch that can be used to link up, manage, and apply security and QOS policies to wireless LAN access points sprinkled across an enterprise. (For more on the debate over what exactly comprises a wireless LAN switch, see Vivato's Switch Bitch.) Extreme’s product "emphasizes security and management," says one source close the company.
Extreme is not the only incumbent systems vendor moving into the crowded 802.11 space, which is both big (see WLAN Worth $1.6B in 2002) and busy (see WLAN Switch Shakeout Looms?). This week Nortel Networks Corp. announced a "security switch," which is expected to ship in June (see Nortel Preps 'Security Switch'). Cisco Systems Inc. (Nasdaq: CSCO) has not announced any plans for a WLAN switch, but observers reckon it's only a matter of time (see Cisco’s LAN Switch: Build or Buy?). Cisco has also invested in a WLAN switch startup called Bandspeed Inc., which is working on technology that enhances the range of 802.11 radios (see Startups Add to Switch Mix).
“I think that longer-term it makes sense for the incumbent switch vendors to... converge wireless and wireline switching,” says Chris Kozup, senior research analyst for global networking strategies at Meta Group Inc.. “It puts them in a better position to win overall. If I’m an IT network manager, I’m sure I’d rather go with an incumbent than a startup who’s not necessarily tried or true in my environment.”
But the trend Extreme is joining is bad news for smaller 802.11 outfits. Incumbent players entering the market have two significant advantages over startups. First, on the business side, they can sell through existing channels and to an installed customer base. Second, on the technology side, they have spent years developing management capabilities. Startups are working from scratch in both regards.
The management angle is particularly important. Part of the 802.11 switch startups' pitch is that they make management simpler by allowing access points to be administered from a single location. But what if the 802.11 switch can’t be managed by the same console that a company uses for the rest of its equipment? Network managers could end up with YAMC (yet another management console) – something of a bitter irony. Buying an 802.11 switch from an incumbent vendor eliminates this risk, because it can be managed using the same console already installed to deal with the vendor's other gear. The incumbent's 802.11 manager is then JANN (just another network node).
In addition to developing a wireless LAN offering, sources say Extreme is also working on new high-speed Ethernet products designed for installation at the core of enterprise and service provider networks. Even fewer details are available on the core products than on the 802.11 offering, and sources had no information about an availability date.
But indications are that Extreme is working not only on a new 10-Gbit/s Ethernet platform, but also a 40-Gbit/s solution. Given that there is currently no standard for 40-Gbit/s Ethernet, it's likely Extreme's new gear will invoke some inverse multiplexing smarts to combine four 10-Gbit/s Ethernet connections into what would be, for now, the ultimate in fat pipes.
Extreme's core developments are clearly part of a strategy to leapfrog its competitors, which are currently fighting it out for 10-Gbit/s Ethernet business. Extreme is facing increasingly bold competition from Cisco and Foundry in the 10-Gbit/s market -– which has of late seen several significant new product announcements, as well as rampant price cutting (see Cisco Takes On 10 GigE Competition and Force10 Slashes 10-GigE Pricing). Riverstone Networks Inc. (Nasdaq: RSTN) is expected to make its own 10-Gbit/s announcement Monday, so keep an eye out.
At least one analyst says she's not surprised to see Extreme eyeing the high-speed core switch market, but predicts challenges. "Demand for bandwidth is not going away, but we won't see much activity [in 40-gig] until the service providers shake themselves out," says Deb Mielke, managing director of Treillage Network Strategies Inc., a consultancy. She doesn't think that will happen until at least the middle of 2004.
Mielke says Extreme's position as an incumbent systems vendor may not help it in the core switch market -- at least among service providers, many of whom still view Extreme as an enterprise play despite its carrier contracts. Indeed, she thinks Extreme risks competition from emerging core startups promising 40-Gbit/s wares, such as PhotonEx Corp. (Providing, of course, the startups survive long enough to compete – see What's Cookin' at Core Startups?)
Extreme could use the financial and market share boosts that sales of new 802.11 and 10/40-Gbit/s technologies would provide. It recently has reported slower demand and falling revenue from sales of its enterprise products, and it’s rumored to be getting set to terminate its content switch line after poor sales (see In Search of... Enterprise Rebound and Extreme to Dump Layer 4-7 Switching?). Expanding its product lines outward (to the edge of enterprise networks) and inward (to the core of enterprise and service provider networks) gives it two ways in which to rally.
— Dan Jones, Senior Editor, Unstrung and Mary Jander, Senior Editor, Light Reading
I’ve read articles on Unstrung about WLAN switch boxes (or shared media hubs), and beamsteering antennas. Vivato claims 33 Mbps. over 100x100 meter office area with 802.11b. Aruba claims 2 Gbps. Capacity with 802.11b.
Here’s the thing, 802.11, without counting any interference from overlapped access points, or hidden nodes, or reduced data rate because of blocked signals, will suffer from increasing numbers of collisions-and-retries as the number of users increases. These collisions cost the network real data throughput, and the congested network phenomenon results.
Vivato claims they can support 150 users in a 100x100 meter office space. That’s about 105,000 square feet. This means that each user occupies about 700 square feet of office space. This means 150 devices congesting the wireles LAN. This means that whenever there isn’t sufficient signal strength between a mobile device and the Vivato antenna, the data rates train in steps from 11 Mbps. down towards 1 Mbps.
Any time a mobile device trains down the data rate, it means that packets take longer to transmit on the network. This means there is less network available for other devices. Which in turn means more collisions and more congestion.
Can an 802.11b network deliver more than 2 Mbps. actual data throughput reliably with more than a handful of users? Where does that leave the Vivato system, perhaps 6 Mbps. (upto 3 simultaneous beams claimed), total throughput for 150 users?
Aruba Networks claims 2 Gpbs. capacity using 802.11b. That translates to just over 180 Access Points.
The question is, can you really use the claimed 2 Gbps.? If an Access Point has a coverage radius of 250 feet for 802.11b operation, then 180 AP’s will cover about 35 million square feet. (PI*R*R) * 180.
Since I don’t think that’s what Aruba Networks had in mind, lets try putting the Access Points close together. If you don’t reduce the power, the AP’s will all interfere with each other, let alone all the devices interfering with each other. If you do reduce the power, then the distance that your mobile device can be from the AP and experience the full 11 Mbps. data rate is severly reduced. Even with reduced power, the nearby mobile devices will act as hidden nodes causing interference, while the co-channel interference from the neighboring AP’s will further reduce the actual data throughput of the wireless LAN.
Lets say you use 180 AP’s in the same 100x100 meter space that the Vivato antenna covers. Essentially every AP can hear every other AP. 802.11b has 3 channels. So lets assume that 60 AP’s interfere with each other on each channel. Lets further assume that mobile devices must roam between these AP’s as they move within this office space, causing hidden node problems, and perhaps some unintentional oscillating between AP’s as the mobile devices try to pick the best AP to associate with. Again, how much actual data throughput does the wireless LAN support? Somewhere between 2 – 6 Mbps.?
Maybe it’s a little better than a Vivato antenna, maybe a little worse. Is either one really much better than a dialup modem in the office environment?
Whats really needed is a wireless LAN technology that can scale to real data throughput of 100’s of Mbps., reaching the equivalent of a Gigabit per second wired network, in a 100x100 meter office space.
A technology that is immune to congestion problems, co-channel interference problems, hidden node problems.
A technology that is resistant to temporary moving obstacles, (such as co-workers), between the mobile device and the Access Point.
A technology that can deliver a sustained network efficiency of better than 85% of the actual maximum wireless LAN maximum data rate.
A technology that can deliver better than 45 Mbps. of real data throughput on a 54 Mbps. WLAN network. A technology that is easy to deploy and doesn’t require extensive site surveys to carefully place Access Points, since neither the number of Access Points nor the number of mobile devices has any effect on the performance and throughput of the network.
What do you call such a technology? Where do you find such a technology?
Corporate WaveNet, Inc.’s TRUEratetm at www.corporatewavenet.com.