Cisco offers WLAN switching
Finally putting pressure on start-ups
After a year of talk, Cisco has finally released long awaited wired/wireless switch capabilities and taken a major step towards an integrated enterprise network.
A key part of the Structured Wireless Aware Network (Swan) roadmap, the new modules - previously codenamed Screaming Eagle and now called WLAN Services Module (WLSM) - add wireless capabilities to the Cisco Ethernet switch. Layer 3 mobility will be added to the Catalyst 6500 Series and will allow users to roam between access points without losing connection.
Screaming Eagle works with the CiscoWorks Wireless LAN Solution Engine (WLSE), a management appliance introduced last year as part of Swan, and the Supervisor 720 module, which includes new Layer 3 mobility management software. Up to 300 access points are supported with up to 16 routing tunnels per AP, and a total of about 6,000 users.
This is the first time that Cisco has produced a wireless switch with centralised intelligence, taking on products from a host of start-ups such as Trapeze and Airespace.
The integrated system is critical if Cisco is to keep its leadership in corporate voice over IP as these systems increasingly move to a mixture of wired and wireless networks, making fast hand-off and quality of service features essential.
Screaming Eagle works by setting up a multipoint GRE (general routing encapsulation) tunnel from the Catalyst 6500 to each AP on the network. However, critics question the efficiency of GRE tunnelling, as it forces packets along a roundabout path known as 'tromboning' that passes through the AP and to multiple switches, before getting to the destination.
But Cisco is presenting its new addition as the first enterprise quality wireless switch, keen to portray the start-ups as providing products most suitable for small and medium-sized organisations, but lacking the functionality, support and financial stability to be trusted in the large corporate network. And of course, it has its massive wired installed base to rely on, looking to milk more revenue from it with wireless extensions.
The product "allows seamless management of what otherwise would be two separate domains", said Larry Birenbaum, senior vice president of Cisco's Ethernet access group. "It is the cap-stone of our Swan announcement a year ago."
However, Cisco has not turned its back on its support for 'fat' access points, which house most of the intelligence and security function in the AP rather than the central switch. This contrasts with the highly centralised approach of the switch start-ups, which rely on dumb APs. Screaming Eagle will work with the Cisco 1100 and 1200 fat APs, though not the older 340s and 350s.
Support for third party APs is not on the agenda, unsurprisingly, given that Cisco has a dominant position in enterprise WLANs and has no desire to let other players in through interoperability. The switch will cost $18,000 in a basic version supporting up to 150 Aironet access points, or $26,000 for the full 300.
The start-ups will now have an even harder job chipping away at Cisco's market share, since their arguments have relied heavily on the giant's lack of centralised management and wired/wireless integration. Their chief remaining argument - apart from technical quibbles over GRE and the possible degradation to hand-off rates - is the lock-in factor resulting from Cisco's refusal to support third party APs.
This will put the spotlight back on the IETF standards body's work on a common specification for dumb APs, which would allow all supporting vendors' switches and clients to intercommunicate, thus reducing customers' risk in choosing equipment from a small vendor.
However, progress towards this standard slowed earlier this year when the proposed Light Weight Access Point Protocol (LWAPP) was renamed and diverted into a seemingly cumbersome process within the IETF.
The LWAPP work was subsumed into a new IETF working group called CAPWAP (Control and Provisioning of Wireless Access Points) and the likely date for completion of a formal document was put back from late this year to mid-2005, giving Cisco more time to solidify its market lead.
The delay is because the mandate for the new working group is initially to study WLAN architecture and topology only - where devices sit and what they are called - without touching the protocols, or how they communicate. This architecture work should take about six months and only then will the group focus on the issues that LWAPP had begun to address.
A few companies say they will continue to support LWAPP in the hope of making it a de facto standard. Prominent among these is Aruba, though it says it will back any final IETF protocol too. Ironically, Cisco also pLANs to support LWAPP - though not CAPWAP - for any functionality that it builds into its upcoming Swan architecture, which will include some measure of switching.
Cisco was one of the original creators of LWAPP, though it backed away from the technology in 2002 as it increasingly set its cap against the thin AP approach. There must be a suspicion that Cisco will seek to hijack LWAPP for its own purposes in order to drive and control standards in this area itself, pre-empting industry body efforts.
© 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.