Recent studies show Linux taking a large and growing share of the global data center market, as well as making incipient gains on the desktop. Traditional IT deployment, however, doesn't tell the whole Linux story - this open source OS is also making impressive inroads in less-visible embedded applications. On this front Linux has come to dominate design-wins in communications, consumer electronics, and other ubiquitous applications, going from upstart to leader in less than four years.

Linux's success in telecommunications and networking infrastructure includes hundreds of shipping systems that do routing, switching, security, wireless and broadband access, billing, and many other data- and voice-centric functions. Emblematic is the success of Carrier Grade Linux (CGL), the specification recommendations created by OSDL members to support the needs of communications carriers and operators. When the CGL specification was first introduced four years ago, it was deemed too "exotic" and specialized; today, CGL capabilities for availability, fault management, performance, and other areas are mainstream in the Linux kernel and software stack. Now in its third revision, the CGL requirements specification informs Linux platform offerings from eight distribution suppliers, powers the product offerings of several dozen networking equipment providers, and underlies voice and data networks around the world. For example, the wireless phone networks of greater Tokyo, much of the United Kingdom, and all of Belgium rely on CGL.

In the last two years, Linux has also enjoyed substantial gains on the "other end of the wire" - garnering impressive design-wins on home networking equipment, digital video set-top boxes, and HDTV sets, and most tantalizingly, on wireless mobile phones. Manufacturers in this 300 million-unit-a-year marketplace are looking to Linux for reasons that mirror Linux's success in the infrastructure: its robustness, performance, security, scalability, and "gold standard" IP networking. Indeed, where mobile OEMs once regarded their wares as hardware-centric, mono-function devices, they today see them as "little servers" - software-dominated platforms for differentiated service delivery.

Analysts estimate that last year between 10 million and 15 million Linux-powered phones entered the marketplace, principally from Matsushita and NEC selling into Japan's NTT DoCoMo advanced wireless network, and also in China from mobile giant Motorola and local suppliers like Datang, Huawei, and ZTE. To augment this organic trend, by further lowering technical and economic barriers to building Linux-based phones, OSDL recently began the Mobile Linux Initiative (MLI). MLI brings together representatives from the whole mobile ecosystem - Linux platform providers and other ISVs, chipset suppliers, handset makers, carriers, and operators - with the goal of accelerating Linux adoption by bridging performance and functional gaps in the Linux kernel and software stack.

Together, CGL and MLI strive to create an unprecedented "end-to-end" deployment of a common platform spanning the gamut of server to client, fixed to mobile, wireline to wireless, and back again. "End-to-end" is appealing in its symmetry and scope, but does such sweeping deployment really confer benefits to networks and their end users? As long as the nodes on a network implement the correct protocols, should we really care about the particulars of the network infrastructure or its clients? Certainly in data infrastructure Linux, legacy Unix and versions of Windows certainly enjoyed and continue to participate in co-deployment on a range of networks, thanks to strong standardization and the overarching need for interoperability. However, designers of embedded software and hardware for client devices face a bewildering array of processor types and embedded RTOSes, competing standards-based implementations, and incompatible middleware. The appeal of an end-to-end, "strategic" platform reflects the needs of OEMs and their customers, the carriers and operators. OEMs see diminishing returns on building and maintaining sui generis proprietary platforms, and equally bad ROI from acquiring, learning, integrating, and maintaining the multiple competing commercial proprietary operating systems that replaced their last-generation in-house creations. Small OEMs lack the resources to build and/or integrate and maintain the millions of lines of code in modern wireless applications. Large OEMs can't afford the overhead of internal fragmentation and seek to leverage the same economies of scale in engineering that they enjoy in manufacturing and distribution. The desire for a unifying strategic platform is especially strong with OEMs whose offerings scale from clients to servers and points in between, like Cisco, Huawei, Motorola, NEC, Siemens, and CE OEMs like Panasonic (Matsushita), Philips, Samsung and Sony.

Carriers and operators (like most end users) strive for "indifference" to the underlying implementation of the devices that support and use their networks. The COTS (Commercial Off-The-Shelf) value proposition dictates that as long as those systems conform to base industry standards and correctly implement key protocols, customers can treat all such equipment as black boxes. However, just as OEMs seek to simplify training, integration, and maintenance, carriers and operators want to unify the management of their network assets. So, while in theory the network is a chain of COTS black boxes, a unifying and open platform represents the infrastructure Holy Grail - a tonic for lagging profits and an elixir to lengthen the life of aging networks and business models.

And Linux, as it turns out, is that Holy Grail OS.

To learn more about how OSDL initiatives are helping to enable global end-to-end infrastructure, visit www.osdl.org.