oreilly.comSafari Books Online.Conferences.


O'Reilly Open Source Convention

Will Money Spoil Open Source?

by Dale Dougherty and David Sims

Michael Tiemann was one of the three founders of Cygnus Corporation, one of the first companies to make its business from servicing open source software. In November 1999, Red Hat agreed to acquire Cygnus in a deal valued at $674 million. When that purchase was finalized in early January, Red Hat appointed Tiemann chief technology officer, replacing Red Hat co-founder Marc Ewing in that position.

O'Reilly Network Publisher Dale Dougherty, Editorial Director David Sims, and Tim O'Reilly sat down with Tiemann a few weeks after the purchase to talk about the integration of Cygnus into Red Hat, the nature of courting developers, and whether money will ruin the open source movement.

listen Listen to this interview:

O'Reilly Open Source Convention Featured Speaker
Michael Tiemann

Red Hat CEO Michael Tiemann is a keynote speaker at the O'Reilly Summit on Open Source Strategies, collocated with the O'Reilly Open Source Convention in San Diego, CA, July 23 - 27, 2001 For more information, visit our conference home page.

• Tiemann co-founded Cygnus Solutions in 1989 with John Gilmore and Gumby.

• The company pioneered open source business models, providing support and custom solutions on open source platforms.

• In November 1999, Red Hat agreed to buy Cygnus in a deal valued at $674 million.

• The two companies see synergies between Cygnus' work in embedded systems and Red Hat's distributions on the desktop.

Dale Dougherty: For those who don't know what Cygnus did before, give us an idea.

Michael Tiemann: Cygnus was the first company to offer the open source business model. We were founded in 1989 and, at that time, we had this thing called free software. Linux would come two years later, but we were selling commercial support for free software and predominantly the GNU tools, GNU C and C++ and the GNU debugger. And, we sold that into initially the Sun market and into HP and Digital and all the different versions of Unix with the theory that one single tool chain across multiple Unix platforms would help the Unix market. Didn't get a lot of mindshare with that, but the embedded systems market wound up really being interested in what we were doing and that became sort of our major contribution to GNU.

Dougherty: So describe that market. What does it consist of?

Tiemann: So, the embedded systems market is often characterized as all the computers that are not on desktops. So, for example, cell phones, car stereos can be embedded systems, microwave ovens, but we tend to do the higher end stuff—datacom and telecom. About 10 of the top 10 telecom and datacom companies use GNU tools from Cygnus to build things like all the Cisco products, the Nortel Meridian phone switch and Passport product. Lucent has three or four major product lines from their Ascend acquisition. And, in fact, even digital audio tape machines are embedded systems.

Dougherty: So, when you say support, in services there's a lot of talk that open source market consists of services, but you guys aren't sitting there answering the phone.

Tiemann: Well, we certainly, we certainly prefer e-mail. But the fact is that a lot of these systems evolved. They're very complex, they live a long time, and people need both innovation and stability and those sound like opposing goals—and often they are. And it's the service and support that makes those two consistent. So, for example, Nortel created the Passport product, which I think was first shipped in about 1994 and since then it's been the best of its class and won several awards and is now responsible for about 20 percent of Nortel's revenue. So it's not a trivial product. But it's got, I think now, over 12 million lines of C++ code for all of its components. We've been evolving our support of C++ along the ANSI C++ standard, but also taking into account the specific requirements of their machines, helping them also port it from one microprocessor design to another. So, there's a lot of stuff that goes on beyond the scenes. Just because you put it into ROM, doesn't mean the code doesn't change.

"For people who want to build embedded Internet appliances with Linux, they couldn't get tool support from Red Hat. They couldn't get Linux from Cygnus, and so they were rolling their own."

David Sims: It seems like a very logical and sort of complementary marriage the two companies then, with Red Hat known for that business model and the desktop and you and the embedded system. Is that kind of the way you see it going forward?

Tiemann: Absolutely. This is obviously a major transaction for Red Hat. Cygnus was the largest open source revenue company, had always been and, in aggregate, had made more profit on open source software than I think probably all of the rest of the open source companies combined. But, as far as this marriage is concerned, it absolutely is complementary. It now gives both Red Hat and Cygnus opportunities that were not available to the two companies separately.

Sims: A little bit more about that. What's the kind of cross-pollinization that you can see now that wasn't possible with just two companies talking together?

Tiemann: Well, a lot of people today are talking about Internet appliances and Internet appliances mean many different things to many different people. To some people, it means a set-top box, to another person it means a single application machine like a web server, you know, which you can install at a highly replicated, very, very dense compact form factor. Well, these Internet appliances actually need embedded systems development tools. Even if they're running Linux, you need to be able to use remote debugging protocols, you need to worry about ROM, you need to worry about memory footprints. So the tools become very important to whether or not your product becomes commercially effective or not. At the same time, the tools are not the whole story. You need a full-featured operating system. For the people who want to go and build embedded Internet appliances with Linux, you need that. They couldn't get tool support from Red Hat, they couldn't get Linux from Cygnus, and so they were rolling their own. Well, as soon as we merged together, the people who were rolling their own, all came and said, "We want the Red Hat brand. We want the Cygnus tools. Let's get going." And the response has just been fantastic.

Dougherty: So, how does this look for the developers? We can look at the Internet IPO stories and all that kind of stuff, but fundamentally developers want to know, is a company like Red Hat going to create more opportunities for them?

Tiemann: That's actually one of our key focus areas. I mean, we've got three major thrusts in our technology roadmap and one of them is ease of deployment. We want to make it possible for Red Hat to be everywhere.

The second is be the developer's best friend. We want to supply development tools for the embedded developer, for the Internet developer, for the server person, for the network administrator, assistant administrator. We want developers to be able to do things that they can not do with any other solution, because the developers are the key to the future and Microsoft figured that out 15 years ago and courted them very, very strongly. We believe that open source is fundamentally more developer-friendly and we want those developers to be rewarded using our solution.

"People who think money is the engine are wrong. Money is just the gas, and the engine is the open source development community."

And then the third piece is really simple and obvious. It's open source and open protocols. Right? We don't want to be competed out of the market because of a proprietary protocol.

Sims: You were talking about courting the developer, I was talking to Paul Everitt of Digital Creations and Zope recently, and he was talking about the amount of work that you have to do to nurture and develop your developer community. It's comparable at least to the amount you would have spent developing the product on your own, it's just that it scales further. And, I wonder what you thought about that. I mean, a lot of people have sort of had the thought that open sourcing is a way to get your development moved off-site.

Tiemann: Well, I think that the people who look at open source as some sort of magic equation where either all of a sudden the world does your work for you or all of a sudden you no longer have to worry about maintaining software. I mean, that's not it at all. Right? Open source is a way of building closer relationships with your customers and better relationships with your developers, but you've still got to come in to work in the morning and you've still got to do something by the end of the day. Back to this IPO question, what do you do with the money? You nurture that infrastructure, you build those bridges, you extend the marketing messages, and you basically make it possible for these developers to participate in a community that's worthwhile, that's moving forward. And that does actually help them see the brighter future that's ahead.

Dougherty: So, you feel like the addition of money to the open source equation can be a positive thing?

Tiemann: Oh, definitely, and I think it's also important to sort of make the following distinction which is, there's the engine and there's the gas. The people who think money is the engine are the people who are wrong. Money is just the gas, and the engine is the open source development community. And the quality of the people who are in that community determines how much horse power this movement's going to have.

Dougherty: It is kind of a promise too, though, seeing the money out there, you say this thing has to develop now, you know, it has to be a certain size, it can't just be anything it wants to be on its own and I think that's a little scary sometimes.

Tiemann: It is a little scary, but I have faith that quality can scale and I mean it's certainly true that many companies, you know, simply try to use scale as a substitute for quality and I think we reject that. We want quality to scale and we want to build and we think that open source is actually a model to help quality scale.

"Gnome enables so many different things. It's just phenomenal."

Dougherty: What open source technologies—Linux is the marquee one—what else do you see out there that's really interesting that is, sort of beyond visibility of a lot of people, but it's starting to spark your interest?

Tiemann: You want me to single one out? OK, here's the thing I love. OK, here's my plug for Gnome. My plug for Gnome is, Gnome enables so many different things it's just phenomenal. Go to and just look at all the different things that people are doing. What's great about Gnome is, it's an infrastructure to support this application environment. It's not just an IDE, it's a whole application infrastructure, but, you know, whether it's charting the positions of the stars to figure out how to point your telescope somewhere or it's doing your taxes, or it's building a drawing program, I just think that there's a whole organic community that has recognized the value of an application environment. So I would say Gnome is a great channel to tune into to see what's happening.

Sims: What do you think about Transmeta?

Dougherty: Although once the mystery's revealed, it's no longer as interesting.

Tiemann: I think there's still a big mystery ahead which is, how are people actually going to use it. Open source would have been a lot less interesting if people didn't constantly try and stretch it and push it in directions that nobody expected. Like Linux going into open systems; now that is interesting. And not just financially to Red Hat. It's very interesting to imagine that we're going to have intelligent devices with an intelligent operating system in them. Finally!

And with respect to Transmeta, I don't expect Transmeta to fill every single one of their potentials themselves. The real question is, can the market find a way to use this in a creative way. Jimi Hendrix completely reinvented music with side effects and things like amplifiers. Maybe we're going to find some interesting side effects in the Transmeta chip that will make it possible to configure computation in ways that we just never thought about in a fixed architecture.

Tim O'Reilly: One of interesting things about Transmeta is Linus, and the fact that it isn't a Linux story. I think that he's demonstrated a kind of leadership there that's similar in some ways to what Tim Berners-Lee did with the Web by saying, I'm not going to just jump on and try to exploit this movement that I started; I'm going to try to go for something else. Now in this particular case, it's in the commercial world, as opposed to choosing between commercial and academia. But I think it demonstrates that same kind of high-mindedness that I think is an important thing in our culture. The fact that people can care about more than the financial rat race is really an important principle. And Linus has said that in a big way by the way he's acted.

In terms of your question earlier for Michael about, to what extent will money distort the open source process and distort the Linux community. I'm not saying that Transmeta may not be a great opportunity for Linus, it may actually turn out to be the way that he makes his fortune. But at the same time, I think the idea that people aren't just motivated by money, and that people are really looking for the right kind of technical challenge is a great message to get out there in the community so that people feel good about not just following the siren song of the dollar.

Tiemann: Another interesting angle on that is that, one of the reasons people join the open source community is that they want their efforts to make a difference. No matter how I cuss and swear at proprietary software, it actually doesn't change its behavior. And this is another interesting thing: How do the markets actually work? No matter how much I want a stock to go up or down, that wanting that desire, that dream actually has no influence whatsoever on the value of that stock. However, if you serve the customer and deliver the product, and you make all those other objectives, the theory is, that will take care of itself. So in a sense, the people who simply look at the numbers without doing the work, that's not the open source way.

Dougherty: Tying the two together, I was thinking of Apple Computer as a company that has had a lot of buzz around it, partically, because it defined a vision of how we want to be. And there's a bit of that around Linux, around open source today Apple Computer's a commercial company, people bought their products and were happy with them. When they've been unhappy with Apple, it's because Apple stopped being that kind of promise to them of an interesting future. When you look at some of the proprietary things, that's part of what people react to: "You're not really giving me a very interesting future to follow."

O'Reilly: The road ahead looks pretty grim.

Tiemann: With respect to Apple, because it's a proprietary system, people have a barometer with only one measurement, which is how are they or not serving the Apple brand. With open source, the spectrum is completely broad. Every computing device, every network component, you can ask the question, "What would open source mean here?" And you can't ask that question about Apple.

O'Reilly: I have to say, you see a fairly constant pressure in the media to put new wine in old bottles, for open source to be just a way of competing with Microsoft, or a way of recreating more cheaply things that we've done before. In fact, those of us who have been around open source have realized that the fundamental driver has been people being able to do for themselves things that commercial vendors weren't providing for them. It's what the leading-edge people do, when nobody's looking, that's going to turn into a big market later. So of course a lot of the really interesting work in the open source community continues to be on the fringes. It's where people are pushing the envelope. It's where people are solving the problems that people don't think are solvable or worth solving and are creating in small ways things that are going to turn into huge markets a few years downstream.

Dale Dougherty is the publisher of the O'Reilly Network. David Sims is the editorial director.

Linux Online Certification

Linux/Unix System Administration Certificate Series
Linux/Unix System Administration Certificate Series — This course series targets both beginning and intermediate Linux/Unix users who want to acquire advanced system administration skills, and to back those skills up with a Certificate from the University of Illinois Office of Continuing Education.

Enroll today!

Linux Resources
  • Linux Online
  • The Linux FAQ
  • Linux Kernel Archives
  • Kernel Traffic

  • Sponsored by: