A Cisco protocol has been chosen as the basis for a standard that will control Wi-Fi access points in enterprise networks -- but analysts say the final standards may be ignored by users.
Cisco's LWAPP protocol has been confirmed as the basis for an eventual standard that could allow users to build enterprise wireless LANs from several vendors' kit. At present, the "thin" access points used in centralized wireless LANs can usually be controlled by central switches from only one vendor.
However, "open" Wi-Fi networks may not make such a big difference. Standard products won't be available for a year, and other vendors have emphasized the difficulty of upgrading to them. In any case, users won't care much, say analysts.
"I think this is less important than it might have been when vendors first pursued it 18-24 months ago or so. Generally it seems that folks go with the same vendor for WLAN switch and for APs," said Infonetics' wireless analyst Richard Webb. This isn't because of compatibility, he said: "They probably get the best deal on APs from their WLAN switch vendor."
To allow multi-vendor switched WLANs, the Internet Engineering Task Force (IETF) has been working for more than a year on the CAPWAP (control and provisioning of wireless access points) protocol. Ironically, the project was kicked off in response to the LWAPP protocol, originally proposed by Airespace as a rallying point for switched wireless vendors, against Cisco.
Last year, Cisco bought into switched WLANs, by acquiring Airespace and LWAPP emerged as a front runner for the CAPWAP standard, in an IETF evaluation report. As a Cisco proposal, however, it now faced competition from SLAPP, a more lightweight approach put forward by Airespace's rivals, Aruba and Trapeze. SLAPP proposed a different approach, sidestepping the idea of a standard to control APs and suggesting a simple protocol that would allow vendors to download firmware to each other's APs.
The LWAPP model was chosen as the basis for the CAPWAP standard by 40 votes to 20. The protocol which defines the functions of the access point (known as the "control plane") will be developed from LWAPP, while components, such as the communications protocol and security specification, will be changed to meet criticism in the review. A working draft will be produced this month, with a final standard in six months, and products in another six months, according to people involved in the process.
Vendors will therefore spend the next year claiming their products will support CAPWAP better and quicker. Both Cisco and the number two in switched WLANs, Aruba, promise to support CAPWAP when it is complete, but Keerti Melkote, marketing vice president of Aruba, reckons that Aruba will actually have a head start.
SLAPP already supports the kind of standard security and tunneling protocols that LWAPP lacks, Melkote claims: "The control plane is more easy to upgrade -- it's a software upgrade," he said. Communications and security components are more basic, but need to run fast with good hardware support, he said, predicting that the eventual standard may pick up the same standards -- GRE and DTLS -- as SLAPP: "Can vendors guarantee an upgrade path, for existing devices?"
Yes, says Cisco's chief technology officer for wireless, Pat Calhoun: "Cisco's infrastructure is programmable, and is capable of handling data format changes." In any case, he predicts that the specific protocols used by SLAPP will be rejected in any case because of "larger implications".
On one level, Cisco may have most to lose, since the best application for CAPWAP would be a firmware upgrade that brings legacy access points under the control of switches. Since Cisco has the biggest installed based of these access points -- under the Aironet brand -- a full embrace of CAPWAP would open its base of unmanaged APs to other vendors.
But it may be that few will bother: "As long as it's not an obvious rip-off," says Webb, "people will probably stick with the vendor they buy their switch from at this stage."
Melkote agrees: "It may be useful at the low end of market. In the enterprise market, we don't see it as that important, although Cisco will try to make a marketing point out of it."
"Time will tell," says Cisco's Cohen. "In the end, radio and RF performance will really matter, which is where Cisco's equipment shines."