Pacific Connection(英語)

Ubuntu: A Keep-it-Simple Linux Distribution Finds Unexpected Success

When people argue the merits of Linux as a desktop operating system, they talk about the open source model, the resistance to viruses, the growing library of applications-and, of course, the low cost. But for ordinary civilians, Linux has lacked one ingredient: an easy-to-install software bundle in which the OS and applications all happily work together. To put it in American parlance, what people outside the priesthood are looking for is a "no-brainer."

Enter Ubuntu, a relatively new distribution that carries the clever marketing tagline: "Linux for human beings." Ubuntu may be the latest passing fad among Linux users, or it may turn out to be a "canonical" distribution-the one everyone pretty much agrees is the entry-level distribution of choice (at least so far as the Linux community every agrees on anything). Either way, Ubuntu is a success story for now. ranks Ubuntu number one in terms of hits-per-day (followed, at this writing, by openSUSE, Fedora, Mandriva, and MEPIS). Because of its easy installation and all-in-one bundle that includes OpenOffice, Ubuntu has captured the mindshare of people looking for desktop alternative to Windows.

Ubuntu has an unlikely history, spawned when a young, wealthy entrepreneur aiming in one direction hit a bulls-eye in another. Ubuntu is the brainchild of South African entrepreneur-turned-philanthropist Mark Shuttleworth. He is wealthy because, while attending the University of Cape Town in 1995, he started Thawte, which became the first Internet certification authority to sell SSL certificates. With the rise of electronic commerce, Shuttleworth's timing was perfect and Verisign purchased Thawte for $575 million.

Shuttleworth used part of his new fortune to purchase a $20 million berth on the Soyuz TM-34, where he flew from Kazakhstan to the International Space Station, spending eight days conducting science experiments. Back on earth, he hatched Ubuntu in 2004 as a way to combine the largess of open source with the needs of developing countries. The word "ubuntu" in both Zulu and Xhosa means "humanity to others" and that seems to have been his intent. But it is hard to evaluate just how successful Shuttleworth has yet been in bringing Linux desktops to poor African villages. Last January, he traveled to Croatia, Pakistan, India, China, Indonesia and Kenya on behalf of the OS.

But even so, Ubuntu's most evident success is the one reflected in the DistroWatch numbers, which show gathering momentum here in the industrial world. There is a server version. There is a Ubuntu billboard advertisement alongside Highway 101 in the Silicon Valley. There is a development, distribution, and promotional organization called Canonical, Ltd. And there are regular, predictable distributions-all free. Canonical says it will ship new releases every six months, support each for 18 months, with daily security fixes. Release names include WartyWarthog, HoaryHedgehog, BreezyBadger, and the latest, version 6-DapperDrake. If you don't have an Internet connection, you can take advantage of what must be the most generous CD distribution policy on the planet: Ubuntu will mail the OS and its bundled software on a disk almost anywhere in the world, free of charge. The only catch: delivery takes six to 10 weeks. This service may seem redundant in an Internet age, but if you are really serious about propagating Linux desktops to the world, it's a nice offer.

Ubuntu split off its server version in October 2005 and it now runs on Sun Fire T1000 and T2000 servers, with support for multi-threading. Another big company is also working with Ubuntu. Google has reportedly created its own internal version of the distribution, rumored to be called Goobuntu. Googling the search term "Goobuntu" shows a scattering of references, along with the unsubstantiated speculation that Google will issue its own distribution. That is not likely.

Three related projects are also underway-all derived from the same Ubuntu source. Kubuntu substitutes KDE for GNOME. Edubuntu is bundled with educational packages aimed at kids. Xubuntu, a less heavily bundled version of the OS, comes closest to Shuttleworth's original goal of a Linux operating environment for developing countries-by supporting older computers with less disk space. Bundled applications on the main Ubuntu distribution CD include OpenOffice, FireFox, an Outlook-like application called Evolution, the Rhythmbox media player and the Totem video player. Beyond the CD, Ubuntu is compatible with a broader "universe" of applications. Ubuntu's licensing policy distinguishes those two categories, plus proprietary binary drivers.

Recent reviews of Ubuntu have ranged from enthusiastic to rave. eWeek Lab gave version 6.06 LTS (which stands for "long term support") its Labs' Analyst's Choice designation, calling the OS its "clear favorite" Linux desktop distribution. "While other popular distributions, such as Fedora and OpenSUSE, also benefit from volunteer, third-party packaging, the Ubuntu community appears to be doing a better job keeping itself organized." One reason for this is the Debian project, from which Ubuntu is based, which is "much more focused on organizing and enabling volunteer packaging efforts than are other Linux distributions-most notably, those of Red Hat."

Comparisons with Red Hat and Debian are inevitable. Last April, Tony Mobily of Free Software Magazine wrote an admittedly controversial editorial predicting that Ubuntu would eventually drive Red Hat out of business, because, he argued, Red Hat has abandoned desktop users in favor of more lucrative corporate installations. (He dismissed Red Hat's own community development project, Fedora, as under-funded and disorganized,) That has left a vacuum for Ubuntu, which Mobily calls "the first desktop GNU/Linux done right. And I mean, really right." Ubuntu, he wrote, has attracted many former Red Hat users, including system administrators-who have now started to use the server version of Ubuntu.

Writing on his blog, Shuttleworth insisted that "Red Hat's demise is not our goal." He said that Red Hat is now "well established in the high-availability enterprise Linux server market, and it will take some years before Ubuntu can make the same claim." In the mean time, Ubuntu will keep doing what it is doing-relying on volunteers and, of course, Shuttleworth's massive cash infusion.

Meanwhile, a controversy has erupted between Ubuntu and the Debian community. Ubuntu is based on Debian GNU/Linux, a distribution created by Ian Murdock in 1993. (The name "Debian" derives from "Debra," Murdock's wife, and "Ian.") Murdock's "manifesto" has been to create a complete open, GNU-licensed distribution "that lives up to the Linux name." The Ubuntu website praises Debian for its large library of free applications, as well as for the large number of hardware architectures supported-more than any other distribution.

According to a CNET report, Murdock has called Ubuntu an excellent distribution, crediting it with growing the Debian universe, but complained that Ubuntu "chose to diverge from Debian rather than to extend the standard Debian core, leading to the inevitable compatibility problems." He wished that Shuttleworth had put its resources into Debian, rather than a derivative distribution.

Shuttleworth duly credits Debian for making Ubuntu possible and hoped the argument would not become Ubuntu versus Debian. But he has argued that Debian's specific goals were unclear. That has meant that the Debian community itself erupts into flame wars over where they are all headed. "Arguments go on forever because one person REALLY wants to see Debian get even more stable on the server, and another person wants to see it get even more cutting edge on the desktop," Shuttleworth wrote on his blog. "One person wants more translation of stable versions of applications, another wants newer versions which are by definition not as well translated. One person wants fewer architectures, another wants the full power of Debian on a small embedded architecture." All of them, he suggested, could be right.

Others outside the organization have put it more bluntly: "Ubuntu's success is a result of both Debian's shortcomings and appalling lack of progress," wrote one commenter on the Shuttleworth's blog. "Ubuntu was exactly what was needed to give Debian a sorely needed kick in the ass."

Sidebar: A Conversation with Mark Shuttleworth

How close has Ubuntu met your expectations, both as an OS for the developing world, and for the industrial world, as well?

If you want to make the most of what Linux has to offer, it's important to look to both. The Linux value proposition is strong in both cases, but it plays to different strengths. In the developing world, there's an enormously strong economic and educational argument in favor of free software. Until now, there hasn't been anybody that was focused on delivering those economic benefits to emerging markets. With Ubuntu, I'm very happy with what we've been able to achieve there.

In the developed world, the focus is more on delivering Linux in the form, packaging and structure that corporations are looking for, particularly on the server. There are much more established companies here-including Red Hat and Novell. I think it will take us some time before we equal them in terms of the breadth of relationships that we have in place.

Were you surprised by Ubuntu's success?
A little bit, yes. I know Linux has a very demanding group of users, folks who know it very well and know what they want. They are also very conscious of the alternatives, so I feel we have to continue to work very hard and to find new things that are exciting, and that energizes us to do good work.
Before you came along there were already a lot of distros. Where do you think you hit a chord?
We were the first to come out focused on making Linux work for the desktop. I believe that it's possible to deliver incrementally better versions of Linux every six months, which surprises people and impresses them with how well Linux can work. Every time we put out a release, we try to take on a few more of the things that people have long considered sacred cows-things that you couldn't do in the free software world. There are some constraints on what we can achieve, but I'm very confident that we can show that in many ways the free software experience can be superior to the proprietary software experience.
For example, people always assume that the installation process for Windows is much easier than that of Linux, and that's not true any more: Linux now has a much easier installation process than Windows. And when you've installed Windows, you then need pick different, mostly proprietary applications, install them and make them work well together. Whereas on the free software side, because free software vendors don't so much compete as collaborate, we're able to make that installation process much more smooth.
Was confining Ubuntu to a single CD part of the "secret sauce?"
Not so much. Rather, it was mandated by a philanthropic efficiency. We do send CDs free of charge to people in remote parts of the world where bandwidth is scarce. As a result, there's a potential benefit to us if we can keep the number of CDs that we have to send down to a minimum. Putting it all on one CD was one of our original higher-level criteria.
For developing countries, just how important is free software?
The really empowering thing is having the ability to modify the software and change it to your particular requirements. We can never get it exactly right for somebody in a remote part of the world with whom we have not had contact. So the best way for us to build a working relationship is for them to take Ubuntu as we ship it, and then modify it to make it better for themselves. And then, for us to take a look at what they've done and perhaps include that in the base platform that we ship in the next release.
What underpins all this is that people are legally entitled to change the software, and we encourage them to do that. That also means that people take our work and put it out under their own brands-and I have no problem with that at all. I think it's healthy and good: they are inevitably adding as much as they might be taking away.
What's next?
My personal challenge is to make the whole thing sustainable, which is a big mountain to climb. As we take on more and more, setting more ambitious goals for development and growing the team, the mountain that I'm trying to climb does get bigger. But I'm confident that we can build something that will be fairly unique-namely, a sustainable organization that pays free software developers to focus on free software development, then make the results freely available. If we can achieve that, that will be worth another shuttle flight.

Sidebar: A Professor Makes the Case for Desktop Linux

Professor Norm Matloff thinks his computer science students should be using Linux on their laptops-so much so that the University of California, Davis instructor has written an introductory beginner guide. "No special sophistication in computers is needed," Matloff writes in the overview. "Any Microsoft Windows user who, say, understands the difference between the C: and A: drives should be able to understand the instructions here and install Linux in an hour's time."

Matloff has himself used a variety of Linux distributions. He currently runs Fedora, installed Mandriva on his daughter's machine, and while he hasn't yet tried Ubuntu, he is recommending it to students based on its reputation, alone. But getting lifelong Windows users to switch has been tough. "On the one hand, the Linux Users Group of Davis is one of the most active LUGs anywhere," Matloff wrote in an email correspondence. "On the other hand, I would say that only about 30 percent of our undergraduate computer science students even have Linux on their machines, and only 15 percent actually use it as their main OS."

Matloff says that one reason has to do with personalities: the instructor teaching beginning programming courses is a huge Windows fan. Another is that most of the programming tools are platform-independent. "Also, college students these days have been using Windows since they were in fourth grade, so it is entrenched in their psyches."

But if Windows is familiar and functional, why should students bother to switch? "I explain it to students in our freshman orientations in very, very pragmatic terms-JOBS!" Matloff points out that familiarity with Linux/Unix is a requirement of many technical employers. He also points out that "by using Linux as their main OS, they will be acting as their own system administrators and thus learning a lot of valuable things that they'll never learn in class. This of course is also true to some extent for Windows, but serious use of Linux tends to be far more "hands on" than Windows, and in any case knowing both systems gives them far more insight. For instance, just having to set up a PC to dual boot both Windows and Linux can give the student a lot of insight as to how the boot process works, what disk partitions are, etc.-which they will never see in coursework.

"I'm sure that these points are not lost on the students, but during those hectic days of starting at a new school, the information they receive during freshman orientation just doesn't have much staying power."

Another factor in favor of Linux is cost. That doesn't much affect U.S. university students, who mostly get Windows bundled on their laptops. But it could matter further down the educational latter, where budgets are usually leaner. Ubuntu's free Edubuntu bundle addresses just this point. But whereas the demand may be greater, the technical sophistication of harried teachers is lower. "When a computing teacher at my local junior high school complained about the expense of using Windows at the school, I suggested using Linux," Matloff recalls. But she replied, 'I don't want to learn anything new.'"

The other obvious use of desktop Linux is in developing nations, the very need that inspired Ubuntu in the first place. Microsoft has responded to the obvious threat of an open source operating system with a pared-down version of XP, which is sold in Thailand, Malaysia, Indonesia, India, Brazil, Mexico and parts of Latin America.

According to Nir Kshetri, an assistant professor at the Bryan School of Business and Economics at the University of North Carolina, Linux can get a foothold in developing countries through a combination of economics and national pride. In China, for example, the cost of Microsoft Windows can equal three to four months of salary, whereas the country has developed a $187 laptop running Linux intended for rural areas. In addition, Microsoft is perceived as an American product whereas Linux is viewed more as a technology of the world. "Whereas people in the U.S. have what is essentially a Microsoft habit, that's not true in China, where they would prefer a domestically-developed operating system." That local version Asianux, co-developed by Red Software Flag, bills itself as the "largest and most rapidly developed Linux vendor in China."