Open Source: Aruba's Hot Commodity

Open-Source Origins
Until recently, WLAN switch makers took a proprietary approach to wireless infrastructure issues; in addition to developing their own switch hardware and software, they would also deploy their own access points. Even so, some WLAN infrastructure software, Melkote said, including certain components within Aruba's switch software, traces its roots back to the open-source world.

Several previous industry efforts sought to build support for fully open, interoperable, access point standards. Efforts such as CAPWAP (Control and Provisioning of Wireless Access Points) and LWAPP (Lightweight Access Point Protocol) would allow an AP to work with any compliant switch, reducing the cost of building a WLAN infrastructure. Such efforts also firmly establish the value of a switched network in its core components--a key objective of Aruba's open-source strategy. While these efforts failed due to the intense competition within the switched WLAN market, Cisco's purchase of Airespace, a leading WLAN startup, illustrated the importance of commoditizing the AP market before Cisco can turn its marketing muscle and installed base against the remaining switch vendors.

Open-Source Vs. "Open" Source
Aruba's open-source initiatives will also have an interesting affect on the WLAN market as a whole: NetGear and Aruba, for example, are already talking publicly about how to position NetGear's traditionally consumer- and small business-focused products for the enterprise market.

On the other side of the table, Trapeze, one of Aruba's biggest competitors (along with Airespace/Cisco) in the switched WLAN market, in early April announced its own "open" initiative to license code and hardware profiles to its partners, enabling them to build limited external management support into their own Atheros-based APs. What effect, if any, Trapeze's initiative will have on its customers' WLAN infrastructure costs remains an open question, especially since Trapeze's partners may be tempted to protect their own margins, "open" or not.

To muddy the waters even more, Aruba and Trapeze have drafted a proposed Secure Light Access Point Protocol (SLAPP), which the companies offered to the Internet Engineering Task Force's CAPWAP working group. This proposal is much simpler than earlier, failed industry standards efforts: It focuses on loading switch software and AP provisioning, which are essentially the same components the two companies addressed with their respective open-source and "open" code licensing schemes.

Even Aruba isn't exactly blazing a new trail with its open-source WLAN offerings. Atheros faces one major competitor: Broadcom, which sells 802.11x chipsets to consumer-focused OEMs, including Linksys, Belkin, Buffalo, and Apple. Yet Broadcom is also an open-source veteran, having adopted an embedded version of GNU/Linux for its chipset reference designs. Many of Broadcom's partners released their own revisions under open-source licenses, and open-source projects have pressed Broadcom's best-selling WRT54G chipset into service for various purposes.

A few other notable, open-source AP code bases play a significant role in the WLAN market, as well. Perhaps the best-known of this group is HostAP, a technology used to negotiate the surprisingly frequent differences between wireless network adapters and access points. HostAP also enables wireless adapters to serve as de facto access points, a function which many Wi-Fi device drivers actually cripple by suppressing the necessary signaling and association behaviors.

One Step--In Which Direction? In the short run, Aruba's open-source shift is likely to have exactly the effect the company envisioned, turning enterprise access-point products, whether hardware or software, into a zero-margin commodity market. In fact, since the players who stayed on the outside are either marginalized (Cisco), mimicking the same approach (Trapeze), or simply irrelevant (Texas Instruments et. al), it would be surprising to see any other outcome. Certainly, Aruba's first open-source steps, however small they may appear, are welcome within the open-source community. That's no guarantee, however that the WLAN developer community will embrace an edge product, such has Aruba's, that still features a highly proprietary set of core components. Aruba has a fine reputation in the industry, but releasing the code for a single, financially insignificant product won't necessarily buy the company anything it doesn't already have today. Ultimately, Aruba's first step, as well as its ultimate goal, will only connect if the company proves itself willing to back up its promises with prompt, bold, and sometimes risky future open-source releases.