How Open Source Grew
The open-source model was invented by Richard Stallman, an exceedingly brilliant MIT computer scientist not known for his love of ideological compromise or corporate profits. In response to the fragmentation of the Unix operating system into proprietary, incompatible dialects, Stallman resigned from MIT in 1984 and started a crusade. He began work on an anti-Unix operating system called GNU, which stands (recursively, of course) for GNU’s Not Unix. He created the Free Software Foundation to distribute that work and the idea of an open-source license to govern it (see “Who Will Own Culture?”).
Although Stallman is rather doctrinaire in his antipathy for business, the world is indebted to him. In 1991, when a 21-year-old Linus Torvalds wrote the original Linux “kernel” – the part of an operating system that controls a computer’ hardware – for his personal computer, Stallman’s ideas informed his decision about how to distribute it. Torvalds is a quietly confident, consummately practical man who has proved to be an impressive leader and manager as well as developer. His creation attracted interest from other programmers, who began to contribute improvements, with Torvalds informally coordinating their work. In the mid-1990s, Linux benefited from two potent forces. The first was the Internet, which enabled electronic software distribution and decentralized collaboration among many programmers working independently. The second force was growing frustration at the limitations imposed by proprietary software vendors – particularly Microsoft and Sun Microsystems.
And so Linux entered commercial use. Its first, and still most successful, niche was Web servers; for at least five years, the majority of the world’s Web servers have used open-source software. Then, several years ago, IBM started to contribute money and programmers to open-source efforts. IBM, Intel, and Dell invested in Red Hat Software, the leading commercial Linux vendor, and Oracle modified its database products to work with Linux. In late 2003, Novell announced its purchase of SuSE, a small German Linux vendor, for more than $200 million. IBM invested $50 million in Novell. IBM, Hewlett-Packard, and Dell began to sell hardware with Linux preinstalled. IBM also supports the Mozilla Foundation, developer of the open-source Firefox browser, and with Intel, HP, and other companies recently created the Open Source Development Labs (OSDL), a consortium promoting the business use of Linux, which has hired Torvalds and other open-source developers.
Now, Linux is running on everything from $80 routers to cell phones to IBM mainframes, and is much more common on desktop PCs. Red Hat is a highly profitable $200 million company growing 50 percent per year, and commercial open-source vendors serve many important software markets. For instance, in databases, there is MySQL, which now has annual revenues of about $20 million, doubling every year. In application servers, there is JBoss, and in Web servers, Covalent.
In the server market, the eventual dominance of Linux seems a foregone conclusion. Michael Tiemann, Red Hat’s vice president for open-source affairs, told me, “Unix is already defeated, and there’s really nothing Microsoft can do either. It’s ours to lose.” Of course, Microsoft, which refused all interview requests for this article, sees things differently. But surveys from IDC indicate that in the server market, Linux revenues are growing at more than 40 percent per year, versus less than 20 percent per year for Windows. Unix, meanwhile, is declining.
Technologically, Windows and proprietary Unix systems, such as Sun’s Solaris, still have some advantages over Linux. But Linux is widely considered to be faster, easier to maintain, and more secure than Windows. As for Solaris, “Sun is very schizophrenic,” observed Tiemann. “They’re dead, too.” Sun recently decided to “open source” Solaris, but most observers feel that that decision has come too late. (Sun, naturally, demurs. “Open sourcing Solaris is a huge step forward,” says Simon Phipps, Sun’s chief technology evangelist.) When I asked Tiemann whether Microsoft could recover control of the server market if Windows went open source, he said no. “Windows is a proprietary, big-company product,” he said. “It isn’t modular or clean enough for outsiders to understand or work on, and it’s too big.”