WLAN switch makers fight for survival
The past two years have seen a host of start-ups taking the gamble that Wi-Fi switches would become a major enterprise networking technology. Now that shipments in something approaching volume are materialising, it's time to separate the winners from the losers.
There are far too many players in the sector, and since the once-sceptical Cisco is also planning WLAN switches, a shake-out is inevitable. There have already been some casualties, such as Legra, and the market going forward, though seeing healthy growth, is likely to support only three or four players besides Cisco. This week, Airespace strengthened its claim to be the start-up most likely to survive, while incumbent market leader Symbol showed some signs of panic, attacking the challenging companies with claims that they are not making real switches at all.
Switch sector growth
In the Wi-Fi market, where rising volumes are being outweighed by a savage price spiral, the switches, which support centralized management and security for a network of 'dumb' access points, are the main source of hope for decent margins. While the home and small office markets are booming in terms of shipments - up 73 per cent to 4.1m units in the third quarter, according to dell'Oro Group - revenue rose by only 21 per cent and is being squeezed still further.
By contrast, the enterprise WLAN sector is growing slowly - up just five per cent in the third quarter, compared to the second, according to Synergy - but margins are higher and switches are the chief driver for an expected growth in 2005 of 35 per cent or more. WLAN switches and 'thin' access points are expected to reach nearly $63m in sales this year, according to Synergy, compared to $20m and $6m, respectively, last year.
Cisco leads the enterprise sector with a massive 43 per cent market share by revenue, and Symbol remains in second place. But for the first time, according to Synergy's quarterly estimate, one of the new breed of switch start-ups has penetrated the top three, with Airespace now in third place, displacing Proxim, which shares fourth place with 3Com, both vendors on around a 4.5 per cent share.
According to Synergy, Airespace made revenues of $18.2m in the third quarter, up 50 per cent and Aruba saw 44 per cent leap to $6.9m. The biggest loser was Extreme, which saw WLAN revenues fall by 55 per cent compared to Q2, to $1.8m. The company has now folded its WLAN division back into the main business.
Airespace forges ahead
Specifically in the switch/lightweight access point market, Symbol leads the pack with 43 per cent share of both technologies, followed by Airespace with 25 per cent of switches and 45 per cent of light APs. Aruba comes in third.
These figures are somewhat surprising since they suggest that all the other light AP makers are only selling 12 per cent between them, despite the fact that light APs are usually specific to their vendor's switches and can't easily be mixed and matched. This suggests that only Symbol and Airespace, so far, are selling to big deployments with large numbers of APs associated with the switch, while others remain confined to the branch office or small business, where fewer APs per switch are required.
The figures position Airespace firmly to be the traditional single start-up that survives independently in any new market. It has built a strong network of OEM deals - with NEC, Alcatel and Nortel - that enable it to ship volumes far beyond start-up capabilities, to large corporate clients. It has also refused to confine itself to the Wi-Fi niche, instead evolving its switches to support multiple technologies including, in the future, WiMAX.
Aruba is also performing well, and could emulate Airespace's success in pushing Proxim aside - or be a takeover target for a larger company such as Cisco. Proxim, though well established, has stumbled in switches before and now seems over-dependent on its voice-over-WLAN venture with Motorola, even as it seeks to expand its broadband wireless activities. Cisco itself will shift the balance significantly as it rolls out its integrated wired/wireless switch strategy under the Swan (Structured Wireless Aware Network) banner.
That leaves Symbol fighting to defend its leadership of the switch sector, and a clutch of start-ups struggling to make any impact at all. Extreme has pulled back while Trapeze, once the most advanced of the new players, is seeking expansion into larger WLANs with a software upgrade to its core product, called Virtual Service Sets.
The new version supports 64 virtual access points (ID channels that are seen by the switch as a single AP) per real AP, instead of the previous two. This is valuable for supporting large or mixed use networks, which can be partitioned to offer different services or levels of access to different users. In Trapeze's enterprise base, this will support the growing trend for companies to segregate access for employees from that of visitors and casual workers. It will also help the switch maker move into the market for large WLANs in conference centers, airports and so on, where rivals such as Chantry Networks are already strong.
As Trapeze fights to fulfil its early promise at last, Symbol, the pioneer of WLAN switches, is seeking to fend off the rise of start-ups by claiming they do not make real switches at all.
It is accusing the start-ups of turning out inferior products masquerading as switches. Its particular grudge is that most of the new companies are using commodity hardware and putting most of the functionality and differentiation into software, which also reduces prices. In Symbol products, most of the clever stuff is in the hardware itself.
Symbol's argument is that it is the only vendor to sell a true light access point, based on its own very thin radios. These devices transfer information directly to the switch as 802.11 data tunnelled through Ethernet and this data is then bridged to Ethernet at the switch. Start-ups such as Airespace and Aruba buy in silicon rather than making their own radios as Symbol does, and their light APs are actually complete consumer-oriented APs with some consumer-oriented features turned off and their intelligence reduced in favour of the switch they connect to.
This makes the start-ups' products bridge managers, not real switches, argues Symbol, defining a real switch as a "concatenation of bridges - as the inventor of the idea, we have the right to define the terms".
This raises many issues, not least the role of standards bodies. We might assume that ongoing standards efforts, notably the IETF's CAPWAP initiative to create a common communication between switch and light AP, should set a new and harmonised set of definitions, but this work is moving slowly and, even within the IETF, the vendors involved are taking contrasting views of how light APs should work and interoperate.
As for the Symbol claims, the start-ups say that, since their APs are more fully configured, they have more intelligence in the AP and can support features such as intrusion detection more easily - even though they run the risk of moving back down the track towards a traditional intelligent AP.
But the fact is that the start-ups are increasingly in the software business, and some have even decided to leave hardware design behind altogether and license their software to other switch makers. Their pattern is likely to win - commodity hardware with differentiated software functionality has tended to be victorious in most mass technology markets, and Symbol may need to adapt to that or risk pricing itself out of the market and leaving it wide open for a Cisco-Airespace double act.
Copyright © 2004, Wireless Watch
Wireless Watch is published by Rethink Research, a London-based IT publishing and consulting firm. This weekly newsletter delivers in-depth analysis and market research of mobile and wireless for business. Subscription details are here.
Sponsored: RAID: End of an era?