Are Wi-Fi switches REALLY about to go open?
Trapeze, Aruba and Airespace all promise to hook up different APs.
Suddenly, all the big players in Wi-Fi switches are promising that users will be able to put networks together from multiple vendors' kit.
- Trapeze has announced an initiative to link other access points, including those from D-Link and 3Com (which re-sells Trapeze switches).
- Aruba has shared the code by which its APs boot up, so other vendors can "Arubacise" APs based on the same hardware: as of 25 March, the code is now posted on SourceForge, and Netgear is the first to promise to do so.
- Cisco is understood to be about to revive LWAPP from its new subsidiary, Airespace, to hook its own access points into Airespace switches. It can't say much, because its acquisition of Airesapce is not yet finalised, but rumours at places like Unstrung, are getting more specific.
But how serious are these proposals?
A break with tradition
Openness goes against the tradition of Wi-Fi switches. For the last couple of years, they have promised to give blanket wireless LAN by managing "thin" access points installed throughout an enterprise, replacing cables with wireless links.
However, unlike the Ethernet cables and switches they want to replace, Wi-Fi switches are proprietary. Buy a switch from one vendor, and you have to buy their access points.
Network managers - even the ones that have blind loyalty to one vendor - don't like to see this kind of blatant lock-in.
In retrospect, it's all a bit obvious. The software is the main part of wireless switching. Ultimately, the hardware - especially the access point - is a commodity, but that isn't so in the early stages of the market.
Network markets evolve in pretty similar ways:
- At first, when there aren't many customers, the vendors make more money - and keep themselves more visible - by putting their badge on everything, and keeping prices high.
- Later on, it makes sense to promise that users can mix and match, in this case putting other's access points on their switches.
- Later still, if we're lucky, market conditions force vendors to deliver on those promises, and actually be open.
We are, roughly speaking, at the second stage. It's really, really important for vendors to pay lip-service to being open. None of them are really sure how necessary it is to actually deliver.
(This refers, of course to the leading start-ups in wireless switching. They are really based around software, implemented on more or less standard hardware. The big player outside of this is Symbol - see Wireless switches - it's hardware versus software. )
A tangled web
For the rest of the WLAN switch industry, this situation leads to some amusing changes of direction, as vendors promote a common method for controlling access points and then change their minds.
Airespace was the first to get publicity for the idea, with its LWAPP protocol. At one stage this was supposed to hurt Cisco, by uniting the other centralised WLAN vendors against Cisco - which was then committed to a distributed architecture with intelligence in the AP.
Somewhat ironically, LWAPP now looks like being just what Cisco needs internally, to glue its legacy Wi-Fi architecture and Airespace's switches together.
Only one other vendor, D-Link, ever promised to deliver an AP with Airespace's LWAPP, and that may have been one of those marriages never intended to be consummated. It went cold well before Cisco bought Airespace, and more than a year after D-Link's promise there's been no sign of any actual LWAPP access point from D-Link.
Funnily enough, D-Link has now jumped into Trapeze's partner alliance, OAPI. The company is making more definite promises this time - an OAPI access point is supposed to arrive in April.
So, right now, everyone has a different standard and, naturally enough, in the good old spirit of industry co-operation, hey are dissing each other's efforts.
"Aruba's announcement is just a marketing story," says Michael Coci, head of technical marketing at Trapeze. "They can link to Accton access points? Well bully for them, that just happens to be the access point they are using themselves. They are just releasing the client software they had on their own AP to anyone with the same hardware."
"It could even backfire," said Coci. "What if a security specialist downloads the software and publishes a security audit?"
"Aruba has decided to shift the support burden to their customers," said Alan Cohen, marketing vice president of Airespace. He also questioned whether Trapeze's effort - to spearhead a de facto standard in the market, would accelerate or slow down the arrival of open WLANs.
Aruba, for its part damns Cisco's use of LWAPP: "Cisco is not committed to interoperability," said Keerti Melkote, Aruba's marketing VP.
What about standards?
Unfortunately, the standards bodies aren't going to sort this out right away. Airespace's LWAPP was proposed as a possible standard for controlling access points 18 months ago, and a final standard has yet to appear.
As it sits at the level of routing software, LWAPP went to the Internet Engineering Task Force (IETF), not the IEEE (which is perhaps just as well, as the IEEE is running out of letters in the alphabet soup of WLAN standards).
The IETF did not simply "approve" LWAPP - it evolved into the CAPWAP working group, which has decided on its objectives, and expects to have a standard (RFC, in IETF terms) in January next year.
Other input to CAPWAP has included CTP (the CAPWAP tunneling protocol), suggested by Chantry and Propagate. Little has been heard of it since - or Chantry - since Siemens bought Chantry at the end of 2004.
So what next?
The positive side of all this is that the problem is finally being addressed - even though the approaches so far may turn out to be lip-service, or limited to a certain "family relationship" of equipment. Switch vendors will find it easy to link to others that share an AP platform, and expedient to link to those with whom they have an OEM or ownership relationship.
As with other systems in the past, the WLAN will open up gradually, reluctantly, and with many false starts.