Despite abundant rivals,, but that's just a means to an end. by extracting revenue from the same server market that Linux leaders Red Hat and Novell specialize in.
Shuttleworth deliberately is taking a different approach from those rivals, though: The free, downloadable version of Ubuntu is the same as the supported, certified version. He hopes to satisfy conservative customers with five-year support plans on versions such as June's; for the leading-edge crowd, versions such as last week's come with 18-month support.
Shuttleworth sold his security firm, Thawte Consulting, to Verisign for $575 million in 2000 and sunk some of the proceeds into a trip into orbit on a Soyuz spacecraft. But his dissatisfaction with the practices of Red Hat and Novell led him back into the computer industry in 2004 with South Africa-based Canonical.
Shuttleworth discussed his agenda with CNET News.com's Stephen Shankland.
You had a lot of options when you were thinking about another start-up. Why did you choose this path?
I tried to avoid it. But I realized that I needed to be busy and active. I'm fascinated by Linux, passionate about Linux. I started making a list of potential projects, and Linux just kept creeping up to the top of the list. I recognized it was going to be a very, very tough competitive environment. It was by far the most audacious (plan), but it was also pissing me off at that time that the established players weren't going to converge on what I saw as the right strategy. In my mind there really was a clear opportunity both commercially and philanthropically.
I think if Red Hat had adopted this strategy--or you could say stuck with this strategy--then I don't think I would be underwriting Ubuntu.
It seems Red Hat's financial success, as opposed to its market share success, finally caught on when they started aggressively monetizing their market share by releasing
I'm not exactly sure. I think we are in a quite a different world today than the one when Red Hat changed their strategy: Linux is much more widely used and it's much more capable as a platform than it was in those days.
Red Hat felt they needed to narrow their scope of their focus to specific market segments, and I think they will continue to dominate those market segments. I think they are a well-run company that makes basically good decisions. They have very good engineering capacity and so on. I don't see that we or anybody else is going to unseat them from that particular position, but the flip side is going to be finding a growth outside of that box. I'm not sure that Wall Street is a warm and cuddly partner in taking on longer-term challenges.
Do you expect most of the money to come from server or desktop support?
In the next three years? From the server.
And after that?
Then it gets more interesting. Who can tell?
So it's open after that?
Yeah. The desktop is very important to us because there's this opportunity to create something new that hasn't existed before. But for us, it's been a way of establishing ourselves. Because other companies had to effectively abandon the desktop, there was an opportunity for us to do something exciting in Linux. If we'd come out with a sever product two years ago, it would have been very difficult to gain this traction that we have now.
So, desktop Linux is a foot in the door to customers.
I've seen a lot of desktop Linux startups fall by the wayside. I've seen Novell and Red Hat try but then step away. It seems you concur it's not a very profitable business.
I agree. I don't think there is money to be made in selling desktop Linux. I don't think the world is looking for another shrink-wrapped proprietary desktop environment. But there is an opportunity to create something great and that you can give away, and we've done that. I feel very good about what we've done on that front and that's opened the door.
Who is your existing customer base?
Right now, it's self-motivated early Linux adopters on the developer and individual side. On the enterprise side, it's people who are managing very large farms of very standardized Linux infrastructure--Web farms, Web hosting companies who provide very scalable outsourced services. Those are the two areas where Ubuntu has a very quick and easy-to-understand value.
Now that you have your long-term support version, Dapper Drake, how will that market change?
We have to continue to deliver the cutting-edge stuff, which is really best encapsulated with our regular six-month releases. That keeps the forward-looking early adopters happy and that's a very essential audience for us. Then on the other side, we have to learn how to offer the things that some of the slower adopters, the more pragmatic adopters, are looking for, which is predictability more than anything else. Predictability around the hardware that we support, the ISVs (independent software vendors) that we work with, the services that we provide them, and so on. One challenge for us is going to be keeping those two quite different groups happy at a time.
Novell and Red Hat both selected the strategy where they have a slow-moving certified version and a fast-moving community version. You have a different strategy, where once every two to three versions, you'll offer long-term support, with the versions in between for including the cutting-edge features.
Exactly. There are a couple differences between our strategy and those of the other players. We would never characterize our six-month releases as beta releases or experimental releases. They are fully supported, and you can buy full contracts on them. They get security updates, which are freely available, so you don't have to subscribe to a network service to get them. They really are genuine, honest-to-goodness, high-quality releases, and they are being deployed in many cases in very high-volume enterprise environments.
We thought the large-scale organizations would tend to want the slower release cycle, but it isn't really working out that way. For example, there is a big debate right now in one of the big Spanish provinces as to whether their next release, which will go on to something like 400,000 desktops, should be based on Dapper or Edgy. That's very interesting, that the latest desktop features are attractive even in a very large deployment.
But you deliberately named it Edgy because you wanted to give the impression that it's full of cutting-edge features. And it has shorter-term, 18-month support. Should a big customer be leery?
It depends on their capacity. For example, Google rolls out Ubuntu to their developer desktops, and they update every six months. So it's not yet entirely clear how people are going to play with this combination of short, fast, controlled six-month releases and then much longer-supported releases.
One thing that means is we're not competing with ourselves. For the organizations which have a free trial community edition and then an enterprise edition, they run into internal conflicts when half of the organization says, "Well, go ahead and use the free thing," and the other half of the organization says, "No, no, no, it's not secure, it's not tested, it's not certified." We think that we would rather avoid that conflict and again really give people what they are looking for.
Certification is a big deal for giving customers the feeling that they can trust their software. Where does Ubuntu stand in the certification process both with the major hardware and software companies?
We're still in the preliminary stages of improving those relationships, but it's absolutely, squarely, centrally, in our sights. We still have a long way to go. I think Red Hat claims somewhere like 400 ISV (certifications), and we possibly have 40. We have not yet announced certification or a relationship with the very largest vendors. Because of the traction that we have in just the sheer number of users, there's a real incentive for ISVs to work on these issues with us.
On the server front, Sun is a good partner of ours. They're really improving their free software story, so I'm quite happy to be associated with them. They have a much better, clearer picture of how they relate to Linux and how they relate to free software. They need to clean up some issues--the Java license and others--but basically I think they're heading in the right direction.
With Sun's UltraSparc T1 Niagara processor, you were the first Linux company to jump on board, in contrast to Novell and Red Hat, which dumped their Sparc versions some years ago. Is there any market interest in Linux on Sparc?
Sun is really onto an interesting idea. They are ahead of the curve. I think they have also done exceptionally well on their Opteron work and their relationship with AMD on that. The relationship between us and Sun allows us access perhaps to jump a couple of steps in this laborious process of building relationships in a customer base.
Do you have support contracts for Linux on Sparc?
We do indeed.
What's the fraction compared to Linux on x86?
It's about 5 percent, perhaps 6 percent of the total subscriptions.
How many subscribers do you have?
I don't think we can say that. We can say that we estimate between 3 and 6 million users on the desktop.
Do you have estimates for the server?
I don't think we track that at this stage.
You guys initially started out with GNOME, and then there's the separate variant Kubuntu with the KDE interface. What's your opinion on that user interface split between KDE and GNOME?
We picked GNOME first because they had a real commitment to usability. KDE had focused on other things. What's really nice about having both is that they get to compete.
On the other hand, from an ISV perspective it forces people to make what's really an unnecessary decision. I wish that there was more compatibility in terms of the licensing of them, and I wish that there was more compatibility in the points of interaction between ISVs and the desktop. Hopefully, that will come over time. Right now we see that KDE has about 30 percent (market share among Ubuntu users) and Gnome has about 60 percent.
Why did you choose to base Ubuntu on Debian versus, for example, something like CentOS where you could perhaps capitalize on the existing Red Hat software and hardware certifications?
Well, it might be just a lack of imagination because I've been a Debian developer since 1996. I would have picked something else if I thought there was another platform that had the same combination of community and engineering quality and focus and diversity in the scale of packages that it provides.
Debian is just unique; it really is in some ways kind of a pinnacle of what the free software world can create. I saw an opportunity to work with that community and take Debian into places where it wouldn't naturally get on its own. There are some folks in the Debian world who say that we've taken from them, and others think we've given them a huge shot in the arm.
Are you trying to improve relations with the non-Ubuntu Debian folks?
It's an area of constant work, and it's also a job that can never be done because community is a dynamic thing. I care a lot about Debian. It's a very important project. It is a project that attracts individualists, so we're never going to get to 100 percent happiness, but we're constantly improving things. For example, we recently agreed to a framework between ourselves and Debian where we mailed them our patches as we make them, instead of publishing them on the Web as we were. That's helped us a little bit in slowly improving relationships. We find some Debian developers want to collaborate and some don't.
The risk if you don't cooperate is you end up forking a lot of packages into potentially incompatible versions, or you work redundantly on the same features.
We do end up diverging from Debian in quite substantial ways, generally because we're adopting new infrastructure. We were really the first to move to Linux kernel 2.6 as a default. We were the first to package X.org instead of XFree86. We do most of the heavy lifting on packaging for GNOME that ultimately does end up in Debian. But it would be wrong to say that we're afraid of diverging. In fact if we never diverged then there'd been no reason for the existence of Ubuntu.