As vice president of IBM Open Systems Development, Dan Frye is responsible for the technical side of IBM's Linux strategy and execution, including the hundreds of full-time programmers within IBM that work directly inside the various Linux communities, enabling IBM's hardware and software for Linux, the relationship with the various Linux communities, technical partnership with the Linux distributors, and responsibility for IBM's participation in various Linux industry initiatives such as The Linux Foundation. Frye recently talked with The IBM LinuxLine about what IBM has learned during its decade of active contribution to the Linux community, the benefits of the "leveraged development model" to customers, and what has surprised him along the way.
LinuxLine: How did you get started with Linux?
Frye: I got started in Linux in 1998, after a varied set of technical and management positions in IBM's HPC software development team- incidentally an early and now spectacularly successful adopter of Linux. I was working on the corporate strategy team responsible for identifying critical technology trends and I wrote a one-page paper recommending the establishment of an IBM strategy for Linux. That one-pager helped launch a formal strategy effort in the fall of 1998 that I co-authored. The resulting actions included broad commitments to support Linux with our hardware and software products, along with the establishment of the Linux Technology Center to improve Linux and build expertise. In addition, that expertise has been transformed into tremendous Linux services competencies. I also now have development responsibilities for elements of cloud computing, small business servers, life sciences applications ... all of which are Linux exploiters (not my doing, just what is happening).
LinuxLine: IBM recently celebrated a decade of active contribution to the Linux community and ecosystem. Why did IBM first get involved?
Frye: For multiple reasons; first, because our customers were beginning to ask for it with our hardware and software. Second, we saw it as opportunity to reach markets and customers that were not traditional IBM customers. Third, we saw a development community and a development process that we felt had legs and we needed to participate in if we wanted measurable success with the first two reasons.
LinuxLine: What has IBM learned as a company as a result?
Frye: A couple of things ... Eric Raymond got it exactly right: "Scratch your itch." That is, don't expect others to do your work for you, join communities and work from the inside to accomplish what you need to get done.
Next, that open source works - the strong, vibrant communities such as Linux, Eclipse, Apache not only produce world-class code, they can be predictably relied upon to produce that code. Internally, we also discovered that some principles of open source development can also be applied to traditional software development (and elsewhere). Tools and techniques for collaboration across large, geographically diverse teams and the philosophy of frequent small updates during development have proven useful beyond Linux and open source, in large part because like the open source community IBM is itself a large, geographically diverse development team.
LinuxLine: What's been the biggest surprise during that time?
Frye: The biggest surprise is a positive one, that the evolution of the Linux development community's processes has matured in advance of industry need. This has helped ensure the second biggest surprise, which is that we never had the broad, sustained quality crisis that many emerging technologies go through between early adoption and enterprise-readiness. The postulation that "Many eyes make all bugs shallow," has absolutely been proven true time and again, from our perspective, and everyone benefits. As to why that is true, well, that's a good debate. I think there's more than one Ph.D. thesis in that question. My own opinion is that it is not the number of eyeballs that counts, but the diversity in the background of the community that is the key. That is, a smaller number of strong developers from diverse backgrounds has more net perspective than a larger number of equivalently strong developers with similar backgrounds. However, I certainly can't prove that. I do think a better understanding of why open source works would be extremely valuable to the industry. Psst ... I'm willing to serve as an industry advisor on a thesis committee. My background in theoretical atomic physics is not all that useful in testing the hypothesis on how developers best work together.
LinuxLine: We've heard IBM talk about the "leveraged development model" of Linux and the broader ecosystem. What does this mean to you, and who benefits from this?
Although IBM is consistently one of the top-contributing companies to Linux and the most active systems vendor contributing to the kernel, what we spend on Linux is a small fraction of what we would spend to enable a single enterprise-class operating system on all of our platforms. The difference in what we spend we can use for other development priorities. Our customers and shareholders both benefit from this leverage.
LinuxLine: As you mention, The Linux Foundation has consistently found IBM to be the top systems vendor contributing to the kernel. What drives such a consistently high contribution rate?
Frye: Because of our ability to combine hardware, software, and services into a single, Linux-based solution, IBM is one of the leading Linux vendors. A constant flow of new products means we always have "more to do" in Linux, much of which is supporting our high-value, differentiated hardware that requires a deep level of integration and support in Linux. Our broad customer base has steadily expanded their use of Linux into new applications areas and new environments. This drives new requirements for Linux and IBM that we help fulfill through our community participation and contributions. Lastly, we think it's fair to do our part.
LinuxLine: Do we have proof that this has driven real benefits for IBM and its clients?
Frye: Yes, we have a substantial, profitable hardware, software, and services Linux business with thousands of happy clients running a variety of workloads, including mission-critical workloads. According to IDC, Linux has been the fastest growing operating systems for some time now, even during the tough economic times of 2008 and 2009-that doesn't happen unless many people are seeing substantial benefits.
LinuxLine: Where do you see Linux going in the next 10 years, and how will IBM help it get there?
Frye: I believe Linux will continue to grow in a variety of places and with a broad spectrum of workloads because it is ideally suited to many ongoing IT trends: virtualization, cloud computing, lights-out management, energy efficiency, cost savings, appliances, and easily managed heterogeneous compute resources, not to mention mobile devices, consumer electronics, social media ... the list goes on. Linux is so widely used - even taken for granted-that it is no longer the "hot story," but if you look at the current "hot stories" I just mentioned, all of them are dominated by or at least have a very strong Linux component. Linux will clearly continue to be integral to the foundation of sophisticated IT capabilities, and will also continue to be an incubator for innovation within and on top of the operating system. Ultimately, the goal is for people to accomplish their computing goals without even knowing about the operating system underneath. Besides, it's cool too!