Airespace

» see airespace.com as it lived, on the wayback machine

~ the last words ~

In 2001 a wireless access point was a small computer. It held its own configuration and made its own decisions about which client to admit, so a hundred of them in a building meant a hundred small computers to administer. Airespace was founded that year in San Jose, under the name Black Storm Networks, on the argument that this was backwards. Put the brains in a controller in a wiring closet and leave the access point as a radio that does what it is told, and the hundred become one thing to manage. There were four founders, two of whom turn up again at the bottom of this page: Pat Calhoun and Bob O'Hara. Storm Ventures and Norwest put up money, with Battery Ventures in later rounds.

The architecture needed a protocol to carry orders from the controller to the radios, so they wrote one. The first Internet-Draft, dated April 2003, is still on the IETF's servers, and the author block reads P. Calhoun, S. Kelly, B. O'Hara, R. Suri, Airespace, with G. Zorn of Cisco and D. Funato of NTT Docomo Labs attached underneath. It was called the Light Weight Access Point Protocol. Everyone shortened it to LWAPP.

For about two years the company was the one to beat. It never had to sell its own brand very hard because it sold other people's: an OEM agreement with NEC in 2003, then Alcatel and Nortel in 2004, each shipping Airespace hardware under their own labels. It was also first to market with location tracking built into the wireless infrastructure, an early version of what later became retail footfall analytics. Headcount ran to around 175 people at 110 Nortech Parkway, San Jose.

Cisco announced in January 2005 that it would pay about $450 million in stock and assumed options. The Cisco quarterly report filed that May records what actually happened, under the deadpan heading of unregistered equity sales: "On March 23, 2005, we issued an aggregate of 22,827,179 shares of our common stock in connection with the acquisition of Airespace, Inc." The final accounting put purchase consideration at $447 million, with $337 million of it booked as goodwill and $95 million as purchased intangibles, the technology itself valued at $78 million over an estimated useful life of four and a half years.

~ the record ~

~ thin and fat ~

The idea won completely. Cisco folded Airespace into what became the Unified Wireless Network, and if you have connected to campus or corporate Wi-Fi at any point in the last twenty years, you have almost certainly talked to a thin access point reporting to a controller somewhere in the building. That arrangement is now so standard that it reads as the only obvious way to build the thing.

In 2003 it was contested. Meru Networks was selling a different answer, and the disagreement stayed live in the trade press long enough that Network Computing ran a Cisco-versus-Meru piece in December 2006. Airespace had been gone more than eighteen months at that point, and the side it had been on was being argued by the company that bought it.

My own first Wi-Fi was a cheap ADSL router under a desk around 2006, one access point with its brains inside it, which is exactly the thing Airespace had been built to argue against. Controllers were a problem for buildings, not for apartments. The nearest I got to this company's world in its lifetime was an internet cafe with the cables stapled along the skirting board.

~ the protocol that lost ~

LWAPP went to the IETF in 2003 and the IETF made a working group out of the problem rather than out of the submission. CAPWAP spent years deciding what a standard access-point control protocol ought to do, treating LWAPP as one input among several, and it published its own answer as RFC 5415 in March 2009. Vendors shipped LWAPP in the meantime, because it existed and their customers had already bought it.

In February 2010 the RFC Editor published LWAPP anyway, as RFC 5412, with a category almost nobody gets: Historic. The status line reads "This document is not an Internet Standards Track specification; it is published for the historical record." An IESG note underneath explains that the text documents the protocol as it stood when it was submitted, and warns that it has not been thoroughly reviewed and may contain errors and omissions.

~ for the historical record ~

Read the author addresses at the back of RFC 5412 and you get the whole arc in a column of postal codes. Pat Calhoun and Rohit Suri are both filed at 170 West Tasman Drive, San Jose, California 95134, and so is their co-author Nancy Cam-Winget. That address is Cisco. Bob O'Hara has no affiliation listed at all, just an email at computer.org. Nearly seven years after the four of them filed the draft from a startup on Nortech Parkway, the document went out under a letterhead that no longer existed except inside their job titles.

The people scattered the way Silicon Valley people do. Bob Friday went on to co-found Mist Systems, which built the next generation of the same idea and sold it to Juniper. The protocol outlasted all of them in one fixed place. RFC 5412 runs to 125 pages, it will sit on rfc-editor.org indefinitely, and it specifies something that never became a standard and ran a very large amount of the world's office Wi-Fi regardless.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »