The LWAPP flap

Lightweight Access Point Protocol, a proposed standard for WLAN interoperability, runs into some static.

A number of wireless LAN switch vendors are pushing a new standard called Lightweight Access Point Protocol that would create interoperability between thin access points and WLAN switches from different vendors, but some key players have failed to jump on the bandwagon.

Alan Cohen, vice president of marketing at Airespace, one of the companies most active in driving the standard forward, says LWAPP addresses one of the major roadblocks to WLAN adoption: the complexities surrounding security, management and deployment.

"Standardization drives adoption," Cohen says. "LWAPP is essentially USB for WLAN [access points] and network devices. USB allows you to plug a printer or a CD burner into a PC, and it connects at a very high speed. With USB in place, the issue of how to connect any new device is taken off the table. This encourages people to create. So when HP comes out with a new photo printer or Apple comes out with the iPod, they just work. The same is true with LWAPP. When you deploy a wireless switch, along with any type of LWAPP-enabled [access point], they will work."

But before LWAPP moves from a proposed protocol to a ratified Internet Engineering Task Force (IETF) standard, it will have to overcome some resistance from Aruba Wireless Networks and Cisco. "It's putting the cart before the horse," says Keerti Melkote, vice president of product management and marketing at Aruba. "Before a protocol is standardized, we must first understand the framework of wireless LANs. What functions belong in the [access point], which ones should reside in the switch? There first needs to be consensus on those points."

LWAPP is a communications protocol for networks that are migrating to "thin" access-point technology. Traditional WLAN products push all traffic handling, authentication, radio frequency management and mobility functions out to individual "fat" access points. The problem is that the access points act in isolation, making it difficult to perform critical functions such as seamless roaming, single sign-on and load balancing.

In small deployments, fat access points are fine. But to be enterprise-class, WLANs must support hundreds of users, and distributing control functions out to the access points forces network managers into a tough trade-off: Do they want robust WLANs supporting many users, or do they want a cost-effective solution? In the fat access point world, they can't have both.

Lightweight Access Point Protocol LWAPP assumes a network configuration that consists of multiple access points connected via Layer 2 (Ethernet) or Layer 3 (IP) to an access router. Today, access routers typically come in the form of WLAN appliances or WLAN switches. The access points are essentially remote RF interfaces that are controlled by the access routers. LWAPP governs how access points and access routers communicate with each other by defining the following activities: • Access point device discovery, information exchange and configuration. • Access point certification and software control. • Packet encapsulation, fragmentation and formatting. • Communications control and management between access point and wireless system device.

Which is where LWAPP comes in. With the arrival of WLAN switching start-ups, there has been a trend toward centralized management, security and control based on thin access points connected to the wired network via a WLAN switch (or a gateway or router). By centralizing intelligence within a WLAN switch, functions such as security, mobility and quality of service (QoS) can be managed across the entire wireless enterprise. However, as more vendors enter the WLAN switching game, the need has emerged for a standardized way for WLAN switches to communicate with access points. Without such a standard, one of the key benefits of thin access point networking - the ability to build multi-vendor WLANs - is lost.

"For enterprises that want to take advantage of new wireless LAN switching architectures, they're left trying to use [SNMP] to communicate with their existing [access points]," says David Passmore, research director at Burton Group. "The problem with SNMP is that many of the features that WLAN switching offers cannot be utilized, such as [radio frequency] management." This leads to relying on one vendor for the switches and access points, and abandoning any access points already deployed in the corporation.

LWAPP centralizes functions that now reside on individual access points, enabling companies to secure and manage their WLAN deployments as a single network, rather than as individual cells. LWAPP would let vendors focus less on switch-to-access-point access protocol, enabling additional innovation in the higher-layer features, such as authentication, packet filtering and policy enforcement. Then any number of higher-layer functions - including encryption, QoS, rogue detection and load balancing - can be centralized for an entire enterprise WLAN.

With LWAPP, a network manager can simply install a WLAN switch in a wiring closet, while relying on LWAPP for access point device discovery. It doesn't matter if the access points are from a different vendor than the switch, as long as the access points also are running LWAPP - the switch automatically will discover and integrate them into the network.

Originally drafted by Airespace and NTT DoCoMo, LWAPP is backed by vendors such as Chantry, Legra, Proxim and Symbol Technologies, with Avaya and Intel recently joining the LWAPP IETF working group. LWAPP is expected to move to a working group within the IETF in the first half of this year, with a final standard expected in 18 to 24 months.

Cisco originally supported LWAPP but then backed away. "We support some of the conceptual ideas that the LWAPP group has proposed, but it's just that it's too early for us to commit to any specific standard right now," says Ron Seide, product line manager in Cisco's wireless networking business. He agrees that there is a general trend toward multi-vendor interoperability. But he says Cisco is more concerned with interoperability from the access point to the client device, rather than from the access point back to the network. He also notes that Cisco is focusing on its own approach to WLAN management and security, its Structured Wireless-Aware Network solution, which was introduced in June.

Trapeze Networks, which originally opposed LWAPP, has since shifted course and decided to back LWAPP, which leaves Aruba as the principal opponent to the protocol. Aruba argues that LWAPP should be set aside in favor of the tunneling protocols already prevalent in networking. "We use a [General Routing Encapsulated] tunnel to communicate between our switches and access points," Melkote says. "GRE is an easy-to-use, 10-year-old standard. Alternatively, vendors could choose IP Security, which is also a proven standard. We favor GRE because it is more lightweight, and that is really one of our concerns with LWAPP. The signaling layer between LWAPP [access points] and switches is very heavy, especially for a so-called lightweight protocol."

Paul DeBeasi, vice president of product management and marketing at Legra Systems, disagrees. "GRE and IPSec are just tunneling protocols," he says. "Neither of them has anything to do with radio technology. Data tunneling is only a piece of what you need. What's really needed is not just taking 802.11 packets and forwarding them to the switch, but a protocol that allows the switch to control the [access points] as well."

"What ultimately will drive the acceptance of a protocol like LWAPP will be, first and foremost, the customers," Burton Group's Passmore says. "Customers want to have choice. They want to be able to pick and choose whatever controllers they want based on relevant features, while retaining the ability to pick and choose access points as well. They are also looking for asset protection or the ability to integrate their existing access points into any new WLAN scheme."