Monthly Columns
 

The Dæmon's Advocate

Copyright © 1999 Wes Peters

Press Forward

Linux has been in the news a lot lately; it has become the darling of the computing press in the past year, and the past season especially. As is usual with the computing press, much of what has been written, both positive and negative, has its roots in the truth, but the story has grown far beyond that original seed.

The discussions about this phenomenon on the FreeBSD Advocates mailing list have ranged from support to confused, angry to irrational. The amazing part is, these discussions often start with a mention of one or more of the BSD systems in the article and then quickly dissolves into a flame-fest over Linux garnering all the publicity.

While it is true that Linux is getting much more press than BSD, the important part to remember is that BSD is getting some press, even if it does mostly follow on the heels of Linux. Is this something to be angry about, or thankful for? Think about that question for a while before you answer.

No Such Thing as Bad Press?

I've also noticed that the mentions of FreeBSD, OpenBSD, and NetBSD in these articles have generally been very flattering. It is almost always stated that the BSD systems have far fewer users than Linux, which is not in doubt, but the single distribution model employed by each of the BSD projects is mentioned in a positive light, as is the licensing model.

This tells us that the BSD community is just about reaching the size to be noted by the popular computing press, at least in the USA, and that what little they know of us is good. We must strive to keep this positive viewpoint. From what I've seen, this shouldn't be a problem. Each of the three open BSD systems continues to improve, to support more hardware, and to meet the needs of more users every day.

Let's take a look at a few recent articles that mention BSD and see what people are saying about us.

An Unlikely Advocate

The first article I'd like to point out is an interview with Alan Cox, one of Linus Torvalds' hand-picked Linux kernel team, in Linux Weekly News. This article has several interesting points. Early in the interview, when asked about the tortuous process of getting the Linux 2.2 kernel release shipped, he points out that the current Linux kernel development model is failing to keep up with demand. He then recommends a new process that sounds amazingly like the core team organization used by each of the BSD projects. [1]

Later in the interview, when asked about competition between Linux and the BSD systems, Mr. Cox points out that Linux has probably helped the BSD projects more than it has hurt them. I agree with him. Once you've convinced people that free, open operating systems are of value, and made them aware that more than one exists, they will naturally begin comparing the offerings. Anyone who brings people in to look around is our friend. He also mentions the support for Linux binaries built into the BSD systems on the x86 architecture, and how this brings the entire suite of Linux applications to BSD. This is a point we should certainly make more of.

Mr. Cox also comments, in an almost backhanded way, on the staying power of the NetBSD and OpenBSD projects. When asked to explain the differences between FreeBSD, OpenBSD, and NetBSD to Linux advocates, I usually start by asking them to explain the difference between Red Hat, Caldera, Suse, and Slackware Linux. I then explain to them the different primary focus each of the three groups has, and how they often and quickly share important pieces like bug reports and fixes, drivers for new or oddball hardware, and even source code management tools. Our diversity makes us stronger.

When asked about the most significant event in the Linux world in 1998, Mr. Cox brings up the issue of getting computer manufacturers to sell machines with the customer's operating system of choice, or at least with no operating system, rather than a Microsoft system pre-installed. He summarizes this OS choice movement with:

It's actually important that it becomes easy to buy a machine without an OS. It would be bad for people like the FreeBSD community if most of the people fighting for OS choice simply said ``OK, now you can have Linux'' and left it at that.
The same obviously applies to all BSDs, and to choice in Linux distributions as well. What good does it do me for Dell to offer the option of Red Hat Linux pre-installed, for a fee, if I want OpenBSD or Slackware?

A Divorce in the Family?

The Gartner Group, one of the most respected of all information technology analysis firms, has been looking at BSD as well. In a recent report titled Divorcing Thin Server Software from the Hardware, J. Staten examines the emerging market for small, task-oriented servers and compares FreeBSD, Linux, embedded Windows NT, and commercial real-time operating systems, often known as RTOSs. Don't try this at home; these are professionals.

This report is very informative, and probably the best short comparison of Linux and FreeBSD I have seen. The factual reporting is quite well done; the performance comparisions between FreeBSD and Linux are quite flattering to FreeBSD. I disagree with the author's conclusions, though. In particular, my Thin Server OS Recommendations would fall along the following lines:

Server Type OS Reason
Intranet/Internet thin server First: BSD
Second: Linux
Both provide fast, stable systems and network stacks at zero cost; BSD protects your intellectual property.
Department/Enterprise Intranet Server First: Linux
Second: BSD
Dead Last: embedded NT
Performance and hardware requirements drive NT to the bottom of the heap. No matter how much IT managers think they want it, they're going to be shocked at the amount of hardware required to support the same number of users on NT, compared to BSD or Linux. We are talking thin servers here, after all.

IT acceptance of Linux is likely to be better for the foreseeable future, mostly due to market and press attention. While this garners it a win in this category for the time being, I don't expect this to hold true for long. BSD continues to provide intellectual property protection.

BSD and Linux score even in interoperability with other systems and management tools, as well as the corporate developer pool most likely familiar with these OSs, since the development model and tools on BSD and Linux are identical. As for corporate standards compliance, this would depend on your corporation, now wouldn't it? BSD is much closer to the standard for MY corporation, which is split about 50/50 between Solaris and SunOS.

Device Control First: RTOS
Second: RTMX/BSD
Third: Linux
A commercial RTOS will provide a smaller, faster system requiring less hardware. BSD or Linux will more easily support emerging standards, and a simpler development model. In the BSD world, RTMX provides true real-time support if needed. BSD licensing provides protection for intellectual property.

The failure of this report to mention NetBSD and OpenBSD is a significant limitation. These systems offer similar advantages over Linux, including intellectual property protection due to the Berkeley licensing model. The discussion of Device Control servers is particularly shortsighted in its failure to mention RTMX, an OpenBSD variant that also provides real time processing support.

An added advantage for each of these systems is current support for the RISC architectures that are popular with embedded developers: SPARC, MIPS, PowerPC, and ARM. While FreeBSD is an excellent system, it limits the choice of CPUs to x86 and Alpha. Neither of these are well suited for small office servers, due to power consumption and heat dissipation issues. The x86 architecture also suffers from price / performance issues.

The Most Unlikely Advocate of All

In an interview in the April 1999 issue of c't, a German [2] computer magazine, Bill Gates himself gave us a plug:

c't: How much competition does Linux really pose to Windows NT? More and more large companies are using Linux. Could that become dangerous for NT?
Gates: Windows NT has always had some competition. FreeBSD has been around for a long time. Apache, free software, has been used by large companies for a long time. Maybe you have only noticed it now. We have had competition from free software for a long time.

Does anyone think Bill Gates considers FreeBSD or Apache a commercial threat to the success of Windows 2000 or IIS? Of course not. There is truth in characterizing FreeBSD and Windows 2000 as competitors, though. My dictionary [3] defines competition as:

  1. the act of seeking or endeavoring to gain that for which another is also striving; rivalry; strife for superiority; as, the competition of two candidates for an office.
  2. a contest; a match.
  3. official participation in organized sport.
  4. opposition, or effective opposition, in a contest or match.
  5. business rivalry; competing for customers or markets.

BSD certainly qualifies as a competitor to Windows 2000 under definitions one and five; Windows 2000 strives for the performance and reliability of BSD while BSD systems strive for the customers and markets of Windows 2000.

Proponents of free markets often talk about competition improving the competitors, but can we see improvement in these competitors? Of course we can. Given the competition that Linux and BSD systems are beginning to provide in the commercial marketplace, Windows 2000 will get better or it will be replaced. Opinions vary widely as to which is more likely to happen, and I'm certain my opinion on this doesn't agree with Mr. Gates' opinion. In return, the developers of the BSD systems, the Linux kernel, and the Linux distributors have all been forced to think about the commercial success of Windows NT and what they can do to ease customers away from NT and into their respective systems. As it has turned out, in many cases, the answer has simply been ``let them know we exist.''

Message in a Bottle?

So, how do we get the message out? Obviously we can't take on the Microsoft marketing machine head-to-head. Even Red Hat, the most commercially successful of all free UNIX distributors, has nowhere near the marketing budget of even a single Microsoft product. How has Linux garnered so much attention in the trade press in the past year, and how can we capitalize on some of that attention?

The number one way to do is it join the press yourself. Write an article about a project you have put together using NetBSD, OpenBSD, or FreeBSD. Write an article about developing applications, or wiring up your home using X10 controllers and a cast off 486 or Macintosh, or whatever you happen to be doing with your BSD system. Contact some trade and popular magazines and find out what they'd like to see in an article.

If you need help writing an article, consult the credits page at Dæmon News and contact a writer whose style you appreciate for help. And don't discount the ``nerd news network,'' including sites like SlashDot and The Register . The Register in particular is very weak on BSD articles.


[1] In the ensuing discussion in the FreeBSD Advocates mailing list, it was suggested that perhaps Mr. Cox et al should get in touch with the FreeBSD and NetBSD CVS managers, and with John Polstra for recommendations on using CVSup. One of the strengths of the BSD development groups is a strong tool-building ethic and that the tools used to build the system are shipped with the system.
[2] English translation provided by Greg Lehey.
[3] Webster's New Universal Unabridged Dictionary, Deluxe Second Edition. One of those wonderful 2,000 page monsters.

Wes Peters, wes@softweyr.com