October 28, 2011

Intel's Dirk Hohndel on 20 Years of Linux

Intel's Dirk Hohndel presented a retrospective on the history of the Linux kernel Thursday at LinuxCon Europe in Prague. Although the 20-year anniversary of Linux has been addressed many times over the course of 2011, Hohndel took his own approach, a personal perspective from a developer present at the beginning of the process and has stayed involved, in one way or other, since.

Hohndel was one of the earliest kernel contributors, and said that he wanted to present his take on the history of the project to provide a perspective that was not focused on the growth of Linux adoption, because for the founders of the kernel, it's primary appeal was as a technical challenge. World domination was an afterthought. In addition, he said, the core kernel team's continued focus on the "next technical hurdle" over the years is actually one of Linux's strengths. That is, they work on the kernel for its own sake. If it wasn't fun for them, it likely wouldn't be a platform for success for anyone else.

In the Beginning...

At the very beginning, Hohndel said, the idea of a new Unix-like kernel project was not novel. "Everybody" was writing their own OS, or so it seemed, with the sudden availability of affordable PC hardware and the GNU utilities, it was easy to get started on such a project for the first time. But Linux was different from the others, he added, because the other hobbyists wanted to retain control over "their" kernels — you were welcome to look, but don't touch, was the message.

Linus was the opposite: he asked for feedback, welcomed patches, and also chose the GPL license that made collaboration over the Internet with remote colleagues possible. In the early days, Linus himself was the number one support provider for Linux — anyone who had a problem could email him, and he would respond.

The timing of the start of the project was critical, but a few other events marked "inflection points" (not quite "turning points") for its future growth. The first of these was the rise of Linux distribution companies in the mid 1990s — they showed for the first time that people could actually make money from Linux and open source software. Primarily they were selling to the inside crowd at first, but it was an important milestone. On the other hand, it also had its downside, in that the distributions were in competition with each other, and their business offices provided the first pressure to differentiate their offering from the others, bringing the first cases of fragmentation to the community.

Hohndel added that he openly accepted responsibility for some of this, having spent time at SUSE. He only wanted to observe that the rise of distributions changed the way the community operated for the first — though definitely not the last — time. That era in Linux also saw the first people leaving the kernel community, as subsystem maintainers left to pursue other projects for the first time.

Avoiding Fragmentation

Linus did a good job of preventing fragmentation from actually splitting the community, he said, in particular preserving the idea that there would be only one kernel. The first embedded Linux systems and the first high-end server users were pulling the project in different directions, but Linus showed that maintaining a single kernel for all of them was not only possible, but better.

The next major inflection point for Linux was the arrival at the end of the 1990s of the first proprietary software that ran on top of Linux distributions. Investment from major IT industry players followed quickly, because for the first time non-Linux companies were beginning to rely on Linux to support their own business. An outgrowth of this era was that kernel development became a profession for developers who had previously done it just for the challenge. From that change came long-term vision for the future of the kernel.

The dot-com bubble (and subsequent bust) were both fortuitous events for Linux, Hohndel argued. The bubble saw start-up companies pour unprecedented amounts of capital into the infrastructure that supported Linux. A lot of venture capital was wasted in those days, he admitted, but the investment meant that businesses had to take Linux seriously. In the bust that followed, and in the 2008 financial crisis, businesses did turn to Linux, out of necessity. By that time, of course, it was more than ready.

Looking forward, Hohndel concluded, there has been talk (even at LinuxCon Europe) speculating on whether or not the "graying" of the core kernel team is good or bad. Hohndel (perhaps unsurprisingly) sees it as good. First, having the same team of subsystem maintainers for several years running has provided Linux with a level of stability over the long term that most commercial products cannot match. But second, it also indicates that the kernel team still enjoys what it does — meeting the next technical challenge, and doing it in an open, "by the geeks, for the geeks" manner.

Click Here!