Other LWN stuff:
 Daily Updates
 Calendar
 Linux Stocks Page
 Book reviews
 Penguin Gallery

 Archives/search
 Use LWN headlines
 Advertise here
 Contact us

Recent features:
- RMS Interview
- 2001 Timeline
- O'Reilly Open Source Conference
- OLS 2001
- Gaël Duval
- Kernel Summit
- Singapore Linux Conference
- djbdns

Here is the permanent site for this page.

See also: last week's LWN.

Leading items and editorials

What it will not have is any representation from the Linux community. Quite a few folks from the community have taken exception to this. Tim Hanson has even announced a protest which will take place outside the event. Free CDs and literature will be handed out in an effort to educate the conference participants about other technologies which should be considered.

LWN asked the organizers about the exclusion of Linux businesses from the conference. The answer we got back was interesting in a couple of regards, and deserves a closer look here.

First of all, the organizers evidently did extend an invitation to Red Hat, which declined to attend. Red Hat apparently sees its best business opportunities in the first world. The conference thus can claim to have not entirely excluded Linux, but it also does not get off the hook quite that easily. Red Hat and Linux are two very different things, and inviting Red Hat is not equivalent to inviting the Linux business community. There may well be other Linux businesses which would have been more interested. Conectiva comes immediately to mind, but there would certainly be several others as well.

The note makes this claim:

We would like to be clear, however, that the Creating Digital Dividends Conference is not focused on specific technologies or operating systems, but rather on new business models that accelerate sustainable development.

What would the organizers think of a business model that:

• Allows third world companies to obtain the software they need to run their businesses without paying large amounts to first-world vendors?

• Eliminates the need to attempt to wipe out pervasive software copying by way of repressive measures pushed by first-world companies?

• Allows third-world hackers to participate in development on an equal footing with the rest?

• Provides the freedom for users to modify the software to fit their particular needs, which may well be different from those of the authors?

Jon 'maddog' Hall has a great story about a third-world cancer screening center which was able to employ a Beowulf cluster to provide near-immediate results from tests. Getting to this clinic was a long and strenuous process for many people; being able to get an answer (and possibly treatment) immediately made the trip worthwhile. It convinced people to go, and saved lives. The Red Escolar project, which is setting up Mexican schools with Linux boxes, is another example of what can be done with free software.

Linux and free software are far from a complete answer to the third world's problems. But a conference that tries to address those problems without Linux is missing a crucial tool. The people who do real work in the third world, however, are increasingly making use of the tools available to them without the need of a conference to tell them what to do. The absence of Linux from this conference is unfortunate, but will not change much in the long run.

Do free software projects need public relations? An often-heard sentiment among KDE developers is that they may have a better desktop, but that the project has taken a number of hits on the public relations front. This idea was made more explicit this week with this KDE Dot News editorial on how to improve KDE's public image.

There is a general consensus that the KDE project, despite its technical superiority among various desktop environments, has had a poor PR record, especially in North America. Now that the release has been delayed a week or so, let's take this opportunity on dot.kde.org to present and share ideas that will help the KDE PR and marketing efforts.

Do free software project need to worry about PR? Part of the mythology of free software is that good code drives out bad, and that the best code wins. So a development project should concentrate its effort on its code, and the rest will take care of itself.

Right?

Of course not. In the business world, simply having the best product is no guarantee of success. The same will certainly hold true in the free software world - especially as the use of free software grows, the stakes get higher, and the amount of money involved increases. There are thousands of development projects out there competing for both users and developers. Good code is a powerful advantage in that competition, but good PR will be important too.

The age of free software project PR may well have had its start at the first LinuxWorld conference in March, 1999. The GNOME project used the event to launch its 1.0 release - and even called a press conference. That move surprised a number of people; after all, press conferences for software releases had not previously been part of the free software development process. That release was part of a well-funded effort to take a project whose code was certainly second-best and make it into a true competitor - and it appears to have worked. Would GNOME be where it is without its PR work?

Development projects - especially large ones - are going to have to put more thought into their PR in the future. One unfortunate consequence of that may be that, in the future, ambitious projects will have a hard time getting off the ground without some sort of corporate sponsorship. That sort of sponsorship is often available, which is an entirely good thing. But free software is supposed to be about what its users want, not what corporations want.

Beowulf 2 from Scyld. Scyld Computing has announced the availability of the second generation of Beowulf software. In this release Scyld is trying to address a number of the difficulties found by users of Beowulf clusters - in particular, the lack of tools to manage clusters and make them appear to be a single system.

This announcement is important for a couple of reasons. Scyld, of course, is the company created by Donald Becker, they guy who first strung together a rack full of Linux systems and called it "Beowulf." He is also, incidentally, the author of a vast number of network drivers in the kernel. Most likely, not even Donald knew what he was setting in motion with that first cluster of his. Beowulfs are now popping up everywhere; for a great many applications they are far more cost effective than the "big iron" supercomputers normally employed for serious number crunching.

Beowulf clusters are not for everybody, however. They remain, to a great extent, a "build it yourself" system involving a fair amount of expertise, time, and duct tape. The users of Beowulf clusters have to be highly aware of how the system is built, and restructure their applications accordingly. Many companies have announced cluster products with nice interfaces, but most of those are oriented toward high-availability web serving. The roots of Beowulf, however, are in hard-core number crunching, and the companies operating in this area (HPTi, Linux Networx, Atipa, and others) have concentrated more on nice hardware.

So the software gap remains. To address this area, Scyld has added a set of cluster configuration and monitoring tools. There is a nice graphical interface and everything. A front-end computer handles administrative tasks, and keeps the whole cluster together. The compute nodes are just that - they even get their operating system from the master system. The whole thing is meant to be easily scalable, so that adding new nodes is a simple task.

As part of this release Scyld is making available a new version of BProc, which is a clustered process management utility, and a thing called Two Kernel Monte which allows substituting a system's kernel "on the fly" without dropping back to the BIOS level.

Those who want to buy a CD with the new code may do so; it's also all available for download from the Scyld web site. The company is clearly planning to make its money on the service side; they offer an array of installation and support plans. The Linux cluster market could well be a large one. The world's appetite for high-end computing continues to grow, and the economics of commodity clusters are persuasive. Donald Becker and company profited little from the first wave of Beowulf clusters; they may do better with the second.

Inside this week's Linux Weekly News:

• Security: New vulnerabilities in ncurses, usermode, and boa. Directory transversal vulnerabilities in web-based scripts.
• Kernel: The Zen of OOM killers; TUX2 patents again.
• Distributions: Coventive arrives in the U.S.; Proposed changes to the Debian Social Contract.
• Development: Python 2.0 release candidate, Microwindows, BIND 9.
• Commerce: E*Trade and the Red Hat IPO; Free Standards Group releases spec; Open Source product announcements.
• Back page: Linux links, this week in Linux history, and letters to the editor

This Week's LWN was brought to you by:

• Jonathan Corbet, Executive Editor
• Elizabeth O. Coolbaugh, Managing Editor

Next: Security