February 2012 CEO Report
Linaro is now receiving industry recognition for its work, both from a purely technical viewpoint, and in helping the member ARM SoC vendors work together on common challenges, while remaining fierce competitors with successful differentiation. This month I’d like to say a little about how we achieve this, and as a result, why Linaro members derive significant value and ROI from their involvement.
Briefly, Linaro’s ability to generate a high ROI comes from each member committing subscription fees and engineering assignees into the company to work on common engineering that can then be leveraged by each member. With 6 members the simplistic equation (ignoring different membership types) is that each member provides 1/6th of the resources, and in return gains all of the engineering output funded by 6 times their own investment. In reality, this only generates an ROI if the work from the Linaro team is needed by each and every member. Does each member really need a third party 120 person software engineering team working on core Linux software? Products were delivered before Linaro existed, and they would still be without Linaro. While Linaro’s output is appearing in an increasing number of high profile products, including those based on Android and Ubuntu, it is hard to say that those products would not be successful without Linaro.
Or is it? Intel has invested hundreds of millions of dollars into open source software over the past two decades. Take this quote from Intel’s Open Source Software brochure: “The industry as a whole is increasingly turning to Open Source as a key driver to rapid, scalable innovation at lower costs. From the start, Intel’s Open Source involvement has been a strategic imperative, enabling end-customers to achieve maximum benefits from the Intel architecture. As a result, the Open Source ecosystem benefits, while the software produced in the Open Source ecosystem is optimized for Intel Inside®.”
How do the ARM vendors address this strategic imperative?
The ecosystem must also be optimized for the ARM architecture. ARM’s business model is substantially different from Intel’s. As an IP provider, ARM’s revenues are orders of magnitude lower. Given the increasing industry importance of open source in general, and specifically to all ARM SoC vendors, an equivalent investment is required to ensure optimization for the ARM architecture. Indeed, given the significant differentiation of ARM SoCs allowed by the ARM architecture, arguably more investment is needed to fully support the ARM partners in the open source world. However, this is clearly beyond ARM’s own resources. Even if it were not, ARM alone could not easily replicate the scope of Intel’s impact on open source, precisely because ARM does not control the SoC features delivered by its customers. While each ARM vendor can and does invest in open source efforts on their own devices, the result to date has been a case of too little too late, insufficient critical mass to influence the upstreams, and substantial fragmentation through multiple vendors implementing similar features in different ways.
This is why Linaro is here and is fundamentally why Linaro is delivering substantial value to its membership.
Linaro’s Not for Profit structure is well suited to attract investment from those ARM partners who understand this the importance of open source software for their customers, and who want to leverage shared investment into Open Source for ARM as opposed to trying to “go it alone”. Linaro is already seeing substantial traction in the upstream community. For example, there is recognition that Linaro’s efforts are, for the first time, contributing to an actual reduction in lines of ARM-architecture specific code in the Linux kernel. The key is that Linaro becomes the place where the multiple ARM licensees can come together to influence the future of open source on ARM. Each company individually, whether small or large, finds it hard to do that – not necessarily because of a lack of desire or resource, but precisely because no single company can speak for its competitors – implementing a piece of ARM specific functionality in one way does not mean that others will do the same thing. The result is that it is hard to upstream core ARM code as an individual ARM SoC vendor. This leads to SoC vendors carrying increasingly large patch sets for their own software functionality. These patch sets must be forward ported into each new version of the Linux kernel, and then into each distribution version for each new product. This is becoming increasingly expensive, delays time to market, and reduces software quality due to lack of common code paths being tested on multiple vendors products. It also ties up big internal software engineering teams, which could be far better utilized delivering differentiated added value.
Linaro’s true value is therefore to those companies who become members. It is not only in the member services that Linaro offers – the landing teams, the LAVA testing platform, the continuously integrated kernel builds on member SoCs and more – but it is also, most importantly, in that shared engineering team. Under the direction of the member-populated Technical Steering Committee, Linaro works on consolidating and optimizing open source software for the ARM architecture, enabling the members to build products out of a single set of upstreamed core software functionality that becomes common across the ARM SoC Linaro members. The engineering output is delivered on our member’s SoCs and becomes part of their own open source software, delivering the ROI from their Linaro membership. Over time this leads to lower maintenance costs, faster time to market, and higher product quality. As Linaro focuses on new ARM technology such as big.LITTLE, multi-core A7 and A15 based products, and then ARM v8, we expect to give our members early competitive advantage in their ability to rapidly deliver differentiated products around a common upstreamed software infrastructure for their SoCs.