Wi-Fi standards still undecided as companies vie to see their technology ahead of the rest

Rethink IT, May, 2005

A new milestone in the quest for an interoperability standard for Wi-Fi access points has been crossed, with the deadline for submissions to the IETF (Internet Engineering Taskforce), which defines standards for Layers 3 and above of the network. The leading proposal has emerged from an unlikely partnership of WLan switch arch-rivals Trapeze and Aruba, with their Secure Light Access Point Protocol (SLAPP) architecture. It will be up against rival draft documents from Panasonic, Siemens/Chantry and Cisco/Airespace.

The standard will be important as it will allow enterprises, for the first time, to mix and match Wi-Fi switches with multivendor 'thin' access points. The freedom from vendor lock-in could bring prices down through increased competition. SLAPP focuses on fairly limited goals in the interests of achieving a standard rapidly, merely allowing multivendor APs and controllers to set up communications in a standard way, but with the actual control signals carried across that link remaining proprietary. This would make it easy for vendors to support the would-be standard quickly, since many protocols would still be supported (in the same way that 802.1x security supports multiple authentication mechanisms). It would also leave plenty of room for suppliers to differentiate their goods by retaining their own signalling methods.

APPLICABLE TO OTHER NETWORKS

The relative simplicity of SLAPP is also important because it would make it easily applicable to other networks such as WiMAX, UltraWideBand and RFID, which will become important as companies move towards systems incorporating multiple radios for different purposes.

However, SLAPP will not be the only proposal on the table for the IETF group (known as CAPWAP). Prominent in the IETF activity to date has been Chantry, a key originator of the CTP (Common Tunnelling Protocol), which is already in use by some vendors and can also be adapted for WiMAX and other networks. Chantry is now owned by Siemens, which has the clout and money to push a technology through a standards process, as does Cisco, new owner of switchmaker Airespace, which helped devise the first would-be standard for thin access points, LWAPP (Light Weight Access Point Protocol). Although submitted to the IETF, time expired and LWAPP was superseded by a new working group called CAPWAP (Control and Provisioning of Wireless Access Points), but Cisco could well breathe new life into its new unit's technology, in a bid to retain control of a key area of the network.

STANDARDS THREAT TO CISCO

Driving standards is important to Cisco since, in networking, it is the only vendor with no burning need to interoperate. Smaller suppliers enhance the appeal of their wares by offering integration with existing Cisco infrastructure, but the giant aims to supply the whole network end-to-end, so that open standards can be a threat unless tightly controlled. Therefore we would expect to see a revival of interest in LWAPP.

The IETF process has been long drawn out in a fast moving market. Last November it reached its first key stage, approving a taxonomy for WLan architectures that detailed the various WLan methods of working in order better to define the requirements for the interoperability protocol. Although most of the focus has been on thin APs, the new taxonomy identified three WLan approaches--autonomous, or fat access points; centralized networks on a switch with thin APs; and distributed or meshed. The likely ratification of the standard will be in mid-2006.

HOW TO DEAL WITH THE MAC LAYER

The main debate has been over whether the media access control (MAC) layer should be separate or not. Splitting the MAC and putting the intelligence on the central switch lowers the cost of the AP and provides better control of the network, argued supporters of this approach, such as Aruba.

The opposite view was represented by Trapeze, which believed keeping the MAC intact allowed for better scaling. Trapeze argued that all WLan chips have embedded encryption, so there is no point turning that off and performing these tasks in the switch, when they could be done more efficiently and cheaply at the AP.

Hence it is a major step forward that these two competing viewpoints have been unified in SLAPP, even though this has been done at the expense of a broader reaching, deeper standard. Compromises of this kind are increasingly important to stop the standards process stalemating altogether, as has happened a couple of times within the IEEE.

Aruba and Trapeze both recently opened up their architectures to third party APs, using different but similar approaches that would both be supported by SLAPP. Trapeze's Open Access Point Initiative (OAPI) enables third parties to make APs compatible with its Mobility Exchange switches, and allows third party software to be integrated into Trapeze Mobility Point APs. OAPI adds a software layer that abstracts an AP's radio technology to erase hardware dependencies.

ARUBA'S AP SOURCECODE

Meanwhile, Aruba is putting its AP sourcecode into the open source process via SourceForge, for the use of third party AP makers that, like Aruba, use the Atheros radio and PowerPC chipset. It will release the sourcecode for the Bins-like boot code that loads at start-up, enabling any AP (including 'fat' ones) to load the code to be centrally managed by an Aruba switch, once it sees one on the network at boot-up.