A little Linux history lesson

Today we’re going to look at Linux. Not as a competitor to Windows and not as a repository of proprietary code (as SCO would have you believe), but as an alternative operating system that might have a place in your network.

Like most overnight sensations, Linux was many years in the making. Its history is generally traced to 1991 but is rooted in the older Minix operating system.

Minix was a Unix-like system developed by Andrew Tannebaum (of the Vrije Universiteit in Holland) for students of his operating systems courses. It was small and somewhat idiosyncratic but did run on an Intel 8086 platform and the source code was published in Dr. Tannebaum’s textbook. The book became an underground bestseller among computer programmers worldwide, one of whom was a 21-year-old Finnish student at the University of Helsinki named Linus Torvalds.

On Aug. 25, 1991, Torvalds announced that he was beginning work on a new system, based on Minix but to be released under the GNU license (we’ll get to GNU next week) so that anyone could acquire the source code, modify it and re-release it.

In 1991, DOS was king (Windows 3 was in beta) in the Intel world. Unix (called Ultrix, SunOS, AIX and other proprietary names) was offered by hardware manufacturers for their own platforms, but not others. There were Unix-like systems for Intel, such as FreeBSD (based on the University of California’s Berkley Software Distribution version of Unix) and the Santa Cruz Operation’s Xenix (which was originally developed by a small company called Microsoft back in 1980). But these were basically niche products, used most by applications vendors to sell “black box” turnkey systems to non-technical commercial users.

Torvalds kept a strong hold on the kernel of Linux while allowing others to develop around it and suggest possible improvements (and even work on them). By the mid-1990s, thousands of hackers (in the best sense of the word) worldwide were contributing to Linux and tens of thousands of users were running it on Intel platforms. This was partially a reaction to Windows (some people just have to get “under the hood” and use the command line) but also a reaction to the seemingly unstoppable march of Microsoft across the face of computing.

When researchers at Los Alamos National Laboratory used Linux to run 68 PCs as a single parallel processing machine in 1996, they were, in effect, building the equivalent of a million dollar supercomputer for about $150,000. That got the attention of enterprise computer users and gave rise to the commercial Linux distributions from Red Hat, SuSE and others. Today, Linux is the fastest growing operating system in history. Read more at https://netfiles.uiuc.edu/rhasan/linux/ about the history of this “guerrilla marketed” (look that one up) operating system.

While many tout Linux as a replacement for Microsoft operating systems, both on the server and on the desktop, most of you aren’t quite ready to take that step. There’s no need to do that, anyway. Heterogeneous networks, with multiple operating systems, are the norm rather than the exception. Look at your own network – is everyone running the same version of the same operating system on their desktop? Also, since Linux was specifically developed for Intel platforms, it runs best on the computers you already own.

The best use of Linux is as a server system running resource intensive applications. Proxy servers, firewalls, Web services, databases, even e-mail (with resource-intensive spam and virus protections) could be cost-effectively run on a Linux platform. Next issue, we’ll look at some specific applications for a Linux box on your Windows network.

CORRECTION: We’d like to clarify that Sender Policy Framework, which was mentioned in last Monday’s newsletter, is now the subject of an IETF working draft.