Weekly Edition
Daily updates
Events Calendar
Book reviews
Penguin Gallery

Surviving the downside, Linuxcare remains focused

An interview with Linuxcare Co-Founder and CTO David L. Sifry

By Michael J. Hammel
Senior Editor, LWN.net

New York - Sitting on the second level of the snack area just outside the main entrance to the LinuxWorld exhibit hall in the Jacob Javits Center, David L. Sifry shows no signs that last years troubles at Linuxcare have left any real weight on his shoulders this year.

"It's very gratifying to know that regardless of all the hype, at the core Linuxcare is founded on pretty sound beliefs."

The CTO and VP of Engineering of Linuxcare has little time these days for the code hacking that got him into Linux. "My hacking level since working at Linuxcare has dropped considerably. That and we had a baby a little over a year ago. No more 20 hour hacking days," he says with a laugh that must make him a favorite target of Santa-suit-wearer seekers.

"I'm a typical geek that found his way to Linux." Like most Linux users, his first move into Linux came as a matter of money. "I didn't have the money to spend $5000 or $10000 on some workstation. I looked around for a while, finally settling on FreeBSD and Linux. Linux was more freewheeling and that was something that I liked. Once I had it running on a box, I just kept it, plus the release cycles were slower with FreeBSD, so I ended up working with Linux."

Sifry is well aware of the developer culture from which the Linux world was born. "When you get into hacking you get into this zone. What tends to happen is that by 9 or 10 in the morning you have the first two cups of coffee and then you get into this zone, where your right brain takes over, and you look up and it's 5PM. You're stomach starts rumbling and you think 'I'm hungry'." His feelings about coding are not uncommon among developers, but things have changed recently. "After having a baby it's much harder to get back into that zone."

That doesn't mean this generally jovial man has had to replace his Jolt cola with espresso completely. He's currently working on a group calendaring project, known as OpenFlock. "We've got great networking and communications," he says of the current state of Linux, "but what we haven't addressed is the concept of group calendaring. I looked around and found a couple of good architectures that weren't being worked on very much. So I decided to do it." But progress has been slow when mixed with the realities of running a company and being a dad.

Last May Linuxcare withdrew it's plans for an IPO, following in August with a round of layoffs. Sifry says to examine how that happened and see what it means you have to go back 9 months prior. "Arthur Tyde, David LaDuke and myself founded Linuxcare based on certain values, things we believed about open source, about building a cool place to work. It was about openness and honesty, about creativity, dignity and integrity. Those were our core values. We had a hot idea to make money on services and not just on open source software like many other companies."

"We managed to get a great group of people working with us: the Rasmus Lerdorf's, the Dave Desrosiers, the Andrew Tridgell's. I hope I don't annoy a bunch of people by not mentioning them - we love them all."

"But fast forward 6 months. We brought on a new CEO [Fernand Sarrat]. Kleiner Perkins felt we needed to do so, to bring in some experienced management. We just got the wrong guy. We got the wrong guy from a culture point of view. We got the wrong guy from a business point of view." Thinking for a moment he adds with another laugh, "We got the wrong guy from a legal point of view," referring to Sarrat's clouded reasons for departure. "We just got the wrong guy. We were too naive. We'd never done a venture funded startup and we knew we had a lot to learn. Thankfully we learned it before the IPO. If we'd have discovered all this after the IPO it could have been disastrous."

"It's very interesting what you learn. I never really thought of honesty as a significant value until it wasn't there. You just assume people will be honest and have integrity. But that's not always true. It comes back to something that my mom, a Holocaust survivor, said: 'they can never take away your education, your experience.' In the end what I've taken away from this is so valuable that all the trouble almost made it worth while to go through it all, to get that education."

But how did the troubles with their ditched CEO, abandoned IPO and layoffs affect the remaining employees, not to mention their customers? Says Sifry, "The good news is that, after we got rid of Fernand, the relationships we'd made with our customers - and our employees - are still there. After all this, we still have all our gurus. Linuxcare has been through a lot in the press, well, you live by the hype and you die by the hype. At the core, it's still a good solid business based on a core set of open source and free software values. Our partners and employees know that. When we did the layoffs it was just right-sizing the company because we weren't going to IPO. We're now somewhere between 150 and 200 people."

Last year Linuxcare went through a round of funding that raised $30 million from corporations and venture capitalists. After the layoffs expectations from the investors may have changed, but Sifry says you can't judge those kinds of expectations. "It depends on the company," he says, clasping his hands as he gets a little more serious. "I can't really speak for them, but for some their expectations are very little. They're just looking at this as a potential hedge for the businesses they do or maybe a long term financial gain. We've always been oversubscribed in terms of investments. We've had more people wanting to invest than we wanted to take, to limit dilution of the company. So we look for good partners that provide significant strategic relationships." Such relationships open doors for Linuxcare by having the investors introduce them to other companies, saying "Linuxcare has some value and, oh by the way, we've invested in them too". It's a way of stating to their customer that the investor has the same beliefs as Linuxcare.

Linuxcare announced in January 2001 a set of contracts with SGI, Motorola, and Maxtor for various services. Sifry says that those contracts required no additional staffing. "For the most part we've taken advantage of the in house talent we currently have."

But all that in-house talent doesn't mean Linuxcare is only interested in growing itself. Last year Linuxcare committed $1 million to the Linux Standards Base (LSB) project, a project aimed at standardizing Linux distributions. Sifry says that committing a million dollars means more than just donations to external organizations. It means doing the work internally and freeing the results.

"We're very big supporters of the LSB and it's umbrella organization, the Free Software Group (FSG)," says Sifry. "A million dollars means a combined budget for items that are LSB officially designated. So that means a certain amount of marketing dollars. It means donations going directly to organizations where that's appropriate. Mostly it means developers salaries. The FSG and other organizations don't necessarily need money, per se, they need developers who are good who are focusing their time on working on the project. We have full time developers working on LSB and part time developers as well. That includes things like man pages or Howto's or other documentation. [Linuxcare's] Christopher Yeoh is one of the lead developers of the test suite used for the LSB. We're still very committed to the LSB, which people have told us they feel will help prevent Linux fragmentation."

As with many Linux companies, people are asking about how to you make money with free software. Linuxcare is a services company, which differs it from vendors like Red Hat, SuSE, and Turbolinux because it doesn't depend on the distribution of software. It also differs Linuxcare from VA Linux, which has a base of hardware to rely on. The question is, how do you scale a services company into the millions of dollars in revenue? Sifry says there are two ways to do that.

"One of those is the EDS route, a body shop, that builds a huge recruiting machine that sends people out and they're going to be onsite making x-number of dollars per hour, and you have to keep them at high utilization rates and deal with all the geographic issues that involves." Linuxcare sees this method as wasteful of resources, primarily their set of Linux gurus.

"The method we've decided on is to use automation as much as possible. The mantra we use at Linuxcare is that we're building a silicon based services business instead of a carbon based one. We don't have to increase the number of people we have to line up with the number of jobs we get. We have a large knowledge base so we can go back and find answers. Even if I don't find the answer I'm looking for, I can find the person I should be talking to. You get this incredible increase of efficiency and scale."

One way Linuxcare addresses this mantra is with their Deltabase product, a subscription based service for identifying the differences between Linux vendor distributions and even between same vendor releases. It also provides news and mailing lists, among other items, that are cross indexed. "We also have managed services, which lets us manage customer systems remotely. Both of these types of services allow us to scale our revenues in a non-linear fashion."

In the February issue of Linux Magazine the Cobalt (ex)CEO Steve DeWitt said that the OS doesn't matter, it's really about the application. Sifry thinks it's a little more complex than that. "Fundamentally I don't disagree. Non-geeks don't buy operating systems. There are people here [at LinuxWorld] that don't know the difference between kernel land and user land, and that's fine. It's a normal maturing process for any technology market. People are interested in solutions, not cool whizbang technologies. How are you going to help them make money or how are you going to help them save money. If there's no real difference between Windows NT and Linux then there's no reason to use Linux. But there are some cases where it's absolutely compelling to use Linux."

Sifry says that developers write code for one of three reasons: to scratch a personal itch, to make money or to help save money. But the business community is looking at this in a more dollars and sense manner. "I'm either going to sell directly to consumers, or in the case of Linuxcare we're selling directly to people who make devices or who write software or to large enterprises. Their needs are very different than the mom and pop consumer market. In our case, we don't have to sell people on Linux. Linux is already the reason people are coming to us."

In most cases customers have already chosen Linux as their platform and need to know if Linux will run on a particular piece of hardware. Linuxcare certifications help determine that a Linux kernel, not a particular distribution, will work on that hardware. Says Sifry, " If I can get it running with the kernel, I can get it running with SuSE, Red Hat, Caldera, Debian or whatever. No distribution maker does [certification] this way. In depth and breadth, our certification (and software testing) goes much further than any distributions. Customers will feel there is a real value in the Linuxcare certification because it really helps their QA teams and the independence of the Linuxcare brand is something they want to associate with."

And is the market free of competition? "It depends on which markets you look at," the Linuxcare CTO smirks. "I'd say the biggest competition isn't a particular company, but the internal IT or engineering department within a company." Sifry feels that most companies are gauging whether to do the work internally or farm it out, but that distribution companies aren't competing for this same business. "We almost never bump into them out there," he says. "To a certain extent we bump into IBM once in a while, but they tend to go into the huge Fortune 500 contracts out there. We're happy if we do the back end work for IBM." And pricing doesn't appear to be a gating factor. "As long as we do a good job they're willing to pay for it."

Recently rumors ran wild on a possible merger between Sifry's Linuxcare and Linux distributor Turbolinux. While nothing official has happened, talks are underway between the two companies. Sifry says he's not closely involved with those talks, but feels the talks are in earnest. "But we've seen no definitive agreement and until that happens it's all just talk. It's happening between the boards." And if a merger happened, what would happen to Linuxcare's distribution independent stance? "We wouldn't be talking to Turbolinux if they didn't feel we could maintain that. There's no indication that independence would change in a merger."

With recent layoffs at SuSE and Turbolinux, everyone wonders if consolidation is a necessary evil in the Linux marketplace. Will there be consolidation? Do distribution vendors need to combine with hardware vendors in order to survive? "Ask me that if and when it all boils out," responds Sifry defiantly. "If Linuxcare stays independent, we'll be fine. We plan on being profitable by the end of the year. If we end up doing [the merger] it's not because there is some overall consolidation going on in the industry. We're doing this because we feel that overall it increases value for everyone."

Sifry is adamant about survival for his company. "Strong companies will survive," he says, "and weak ones won't. Linuxcare has 30% of the market in services for Linux right now. That's not bad considering HP, IBM, VA and others are in it. Is consolidation necessary? I don't think so. Will it happen? Probably. There are more than short term business reasons to do so, to make companies faster, smarter, quicker."

For now, Linuxcare is managing to be fast enough, smart enough, and quick enough on it's own.