Report from the first Desktop Linux Conference

Email.Email weblog link
Blog this.Blog this

Andy Oram
Nov. 11, 2003 05:12 AM

Atom feed for this author. RSS 1.0 feed for this author. RSS 2.0 feed for this author.


It was a lucky stroke of fate that brought the first Desktop Linux Conference on November 10, a time of some distress among Linux proponents. In the past two months we have had to assimilate Red Hat's abrupt change in licensing, which reduced its consumer presence (although, as VP Brian Stevens pointed out at the conference, Red Hat Professional Workstation is still available, albeit at a higher price than earlier offerings). Rubbing in the message even further was the blunt declaration a week ago by Red Hat CEO Matthew Szulik that Linux was not ready for ordinary home users: "I would say that for the consumer market place, Windows probably continues to be the right product line."

The purchase of SuSE by Novell offered a bounce in the news, but we still don't know what difference that will make.

In this context, a spanking new and dynamic Desktop Linux Conference from the Desktop Linux Consortium can remind us that Linux adoption is on a strong upward trend. The central lesson did not depart much, actually, from Szulik's assessment. Nat Friedman, cofounder of Ximian, pointed out in his keynote that it's not a question of "when will Linux be ready for the desktop," as if some key feature will fall into place and cause a "big bang" that leads to instant acceptance. Instead, Linux will advance steadily in the desktop space, as it has in the server space. Sam Docknevich, Linux and Grid Services Executive at IBM, pointed out that Linux penetration on the desktop is actually proceeding faster than its penetration among servers. For the past couple years, desktop adoption has been growing at 44% a year.

Friedman, Docknevich, and others gave similar messages: that one doesn't get far by indiscriminately pushing Linux everywhere, but must determine what environments and users are ready for Linux. The speakers even charted a similar roadmap for the stages of adoption.

First is the kiosk: point-of-sale terminals and other systems with extremely limited applications, where programmer control over user activity is fairly complete. Linux is already making big inroads in that market.

Second are technical users, who often use Sun or SGI workstations and are now moving to Linux. Learning Linux is easy for them, and many applications are already ported.

Successive trends are barely starting, but will come in time. And probably sooner than the five years or so that many analysts predict.

The next stage will bring Linux to workers who need computers, but for fairly basic operations. Their usage does not go beyond email, Web browsing, moderate use of word processors or other office applications, and perhaps one or two custom applications that a company can port to Linux. These people can often use Linux with Evolution, OpenOffice, and Mozilla without even realizing that they aren't on Windows. A thin client solution such as Linux Terminal Server Project will make migration particularly easy, and maximize financial savings.

Last, and most difficult to conquer, are the advanced knowledge workers. These people often carry out sophisticated tasks with specialized features of proprietary software that aren't reproduced in free software equivalents. The path for them may involve an emulator such as CodeWeavers (based on Wine) or NeTraverse, both of which were introduced a full session of about 35 attendees. Jeremy White of CodeWeavers explained that he considers Wine "a bridge to a Linux desktop."

Other presentations at the conference covered other projects that are currently driving Linux adoption or will be potentially significant in the future: LTSP, SELinux, OpenSSL, GNOME, KDE, and

But the conference atmosphere did not always consist of holding hands and singing hosannah. Significant disagreements were aired. Several participants criticized Red Hat's new licensing policies and suggested either that they violated the spirit of the free software movement (because of restrictions placed on customers) or were simply part of an outdated and doomed pricing strategy. "Selling Linux by the seat," said Desktop Linux Consortium spokesperson Bruce Perens in his keynote, "may not be healthy for Linux." And Nat Friedman declared, "Per-seat desktop licensing is dead." He suggested we "fundamentally alter the economics" through thin client solutions such as LTSP.

Posing a direct challenge to the Red Hat strategy, Perens announced a "user-driven Linux" initiative. Here, major customers would pool their money and put it into creating an alternative enterprise Linux that would be distributed freely, instead of paying licenses. While he acknowledged that Red Hat's opening up of Fedora met some of the same needs as the GNU project's open Debian project, he said Fedora "just feels too much like putting money in Red Hat's pocket. I'd rather have it go to the users."

Brian Stevens of Red Hat, when the time came for his session, got to present an explanation and justification of Red Hat's current strategy, as well as showing how their development process supports a reliable and easy-to-maintain set of products.

He cited the Open Source Architecture announced by Red Hat in September 2003, which built on an announcement made a month earlier at LinuxWorld and reported in my weblog from there.

Red Hat is moving "up the stack," including applications and other components, "building full enterprise solutions." Examples of the customer concerns Red Hat has to address include how to do single sign-on, and making sure the operating system has the resources needed to run a high-availability database cluster. Perhaps predictably, Red Hat is sounding less like a "Linux company" and more like a "solutions company," where it will compete with a different set of vendors.

Stevens also pointed out that the move from high-end Unix systems to Linux on cheap Intel chips has caused the number of systems to proliferate. Customers tend to run a single server on each system. And the use of many systems creates a management problem.

Perens presented a talk on the SCO situation (which I couldn't get to) and Tom Adelstein, of Project Leopard and the Open Government Interoperabilty Project, described an initiative to define standards for critical government software and disseminate open source solutions. Right now, governments everywhere are struggling with old software that vendors no longer support. Their systems are out of sync with their own standards, and therefore unable to communicate with the systems in other agencies to carry out critical tasks such as exchanging criminal records. With Project Leopard, not only will governments save millions of dollars, but for the first time they will have software that actually interoperates and conforms to the standards they wrote.

All in all, the Desktop Linux Conference is a big step forward. Thanks go to the Desktop Linux Consortium, who announced that they will soon be offering membership to individuals, and to Boston University for providing an excellent facility and support. I will post a follow-up analysis in a couple days.

Andy Oram is an editor for O'Reilly Media, specializing in Linux and free software books, and a member of Computer Professionals for Social Responsibility. His web site is