Volume 7, Number 1, Pages 6 + 37-38

Apache's Free Software Warriors

by Andrew Leonard

This article first appeared in Salon, an online magazine, at http://www.salonmagazine.com. An online version remains in the SALON archives.
Reprinted with permission.

The most popular Web server software on the Net isn't made by Microsoft. Or Netscape. It isn't made by any traditional company at all. How'd that happen?

A red feather resting against the words "Powered by Apache": On the Web, the logo is everywhere. And for good reason- the Apache Web server, a piece of software that transforms ordinary computers into sites on the World Wide Web, is by far the most popular choice for Webmasters everywhere.

That fact alone is extraordinary. The Apache server is not a commercial product- it is "freeware" hacked together by a loose coalition of software developers stealing spare time from their demanding day jobs. And yet, in a market niche targeted by both Microsoft and Netscape as a crucial steppingstone to future profits, Apache dominates- it is installed on almost half of all publicly accessible Web servers. And not just because the price is right, say Apache fans, but because it is a better product than its commercial competitors faster, more reliable, more up-to-date.

Market share isn't the only yardstick by which to judge Apache. Its success pays tribute to the utopian ideals of the whole "free software movement"- and testifies to the enduring vigor of the Internet's cooperative, distributed approach to solving problems. Never mind the buzzword-hyping marketroids endlessly pushing the latest shrink-wrapped solution to all software problems. The Net offers a better way, as even Apache's competitors will concede- off the record.

"We think the Apache guys are way cool," says one Microsoft employee who didn't want his name used. "We think they're at the heart and soul of what the Internet is all about."

Ironically, despite Apache's prominence in a noncommercial movement, it owes its birth in part to one of the first capitalist encroachments upon the Web.

Back in late 1994, Brian Behlendorf, a young programmer fresh out of UC-Berkeley's computer science department, was entrusted with the task of setting up the Web site for HotWired, Wired Magazine's first foray into online publishing. One of HotWired's initial requirements was a registration system that required passwords- a controversial concept at the time.

In those days, most Webmasters depended on a Web server program developed at the University of Illinois' National Center for Supercomputing Applications (also the birthplace of the groundbreaking Mosaic Web browser). But the NCSA Web server couldn't handle password authentication on the scale that HotWired needed.

Luckily, the NCSA server was in the public domain, which meant that the source code was free to all comers. So Behlendorf exercised the hacker prerogative: He wrote some new code, a "patch" to the NCSA Web server that took care of the problem.

He wasn't the only clever programmer rummaging through the NCSA code that winter. All across the exploding Web, other Webmasters were finding it necessary to take matters into their own keyboards. The original code had been left to gather virtual dust when its primary programmer, University of Illinois student Rob McCool, had been scooped up (along with Marc Andreessen and Lynx author Eric Bina) by a little company in Silicon Valley named Netscape. Meanwhile, the Web refused to stop growing- and kept creating new problems for Web servers to cope with. Patches proliferated in an uncoordinated, hopscotch manner.

Behlendorf is a fervent believer in the possibilities for cooperative software development on the Internet. He began to contact the other programmers and suggest that they join forces in rewriting the NCSA code. At first, the developers called their stitched-together solution by the joke name "A PAtCHy server." But by February 1995, they had completely rewritten the NCSA program from top to bottom and formalized their cooperation under the name "the Apache Group." Ever since, collaboration across the Net has been the Apache watchword.

"Everything that we've done has been in a distributed manner," says Behlendorf. "There hasn't been one company, one person, or one group of people that's controlled development."

Today, according to Netcraft, a British network consultancy, 49.7 percent of all publicly accessible Web servers on the Internet run Apache- and the number keeps growing. (The program runs mostly on Unix operating systems, though a new version is now available for Windows NT.) Apache's proficiency at allowing a single machine to host thousands of different virtual Web domains has given it a commanding share of the Internet Service Provider market. Yahoo runs Apache, as does the Internet Movie Database and, not surprisingly, HotWired itself- although Behlendorf, now the chief technical officer at the Web production studio Organic Online, is long gone.

Apache's success is worn like a badge of honor by free software enthusiasts, and touted endlessly as proof that the Internet hasn't been utterly overwhelmed by voracious commercial software titans like Microsoft.

"Apache is one of our great success stories," says free software advocate Eric Raymond, author of "The Hacker's Dictionary." "Everybody thinks so. And the Apache people see themselves, with justification, as exemplars of free-software culture."

But just what is free software "culture"? The free software community is large and diverse, not to mention packed chock-full of acronyms only a geek could love- GNU, Linux, FreeBSD, Perl. It is riven by religious schisms and periodically explodes into flame wars over technicalities that might strike the outsider as enormously picayune. Still, despite their differences, free software developers share the belief that their movement is not only the Internet's past but its future.

"The Linux guys and the Apache guys and the Perl guys and the W3C and the people hacking Internet standards at the IETF all identify with each other, with the free software community in general, as much or more than they do with individual projects," says Raymond. "I'm not an Apache developer, but I nevertheless think of their success as 'ours' rather than 'theirs,' and they feel the same way whenever Linux or GCC scores another design win."

The first thing to understand about free software is that it is not necessarily free. As legendary MIT hacker Richard Stallman, the founder of the Free Software Foundation and primary coordinator of the GNU ("GNU is Not Unix") Project, likes to say, "Free software is a matter of freedom, not price- the freedom to modify the software, redistribute the software and release improved versions of the software."

Such "freedom" basically means the ability to access- and change- a program's source code. The typical software company guards its source code like a pack of pit bulls standing watch over crown jewels. Source code is considered the core intellectual property upon which the software industry is built. For example, Microsoft's Internet Explorer Web browser may be free to consumers, but it is not "free software" - the source code is under lock and key.

Source code availability is the great selling point, so to speak, for free software (and, indeed, Behlendorf would rather use the phrase "source-code available software" than "free software"). The point is, if something goes wrong, you don't need to wait around for Microsoft or Netscape to answer your support calls. If you discover a bug or you need some new function, you can just roll up your sleeves and start mucking about with the code yourself. And more often than not, someone else already encountered the same problem and posted a fix to the Net.

"Having access to the code and being able to modify it is pretty key," says David Filo, co-founder of Yahoo. "It's just a lot easier when you have total control over it."

However, there are different kinds of total control. Stallman's GNU software is protected by a licensing system that requires anyone who distributes modified versions of GNU software to distribute the modified source code as well. This ensures that each future generation of the software will always remain open.

The Apache Web server, on the other hand, is protected under a different, less restrictive licensing system that allows future users to do whatever they want (as long as they include the Apache Group's copyright notice). Already, several companies are reported to be turning a profit with commercial versions of Apache, most notably C2Net, an Oakland, Calif., software company that sells Stronghold, an Apache Web server incorporating public key cryptography.

Behlendorf goes so far as to argue that Apache's licensing is "more free" because it places fewer restrictions on future users. And he believes that this has helped the spread of Apache. But such an approach is anathema to Stallman. If source code availability isn't permanently protected, he argues, then the software can ultimately be perverted by commercial pressures.

Stallman believes the right to free software is "inalienable." To him, it's a moral issue; he doesn't care whether free software produces better code or is embraced by consumers.

"I don't judge these things in terms of material benefit," says Stallman. "My fundamental reason for having a free software community is so that I can continue to use computers without ever doing something wrong- without ever promising that I won't share with you. Even if most computer users don't choose to be in this community, I can be in it. It's not a matter of whether one philosophy or another wins the whole world. It's not an all or nothing thing."