Red Hat: Linux won't rule

Bob Young explains why converting Unix customers is easy and why Linux will never replace Windows on the desktop.
Written by Matthew Broersma, Contributor
Q&A Leading Linux vendor Red Hat said at last month's earnings announcement that it would be emphasizing its focus on two key areas: replacing legacy Unix systems, and expanding its presence in the embedded market. That strategy is paying off so far: despite the continued world economic slump, Red Hat showed a slight profit for the third quarter of its 2001 financial year.

In a conversation with ZDNet UK, Red Hat Chairman Bob Young explains why converting Unix customers is easy; why open source will win the day on the Internet; and why Linux will never replace Windows on the desktop.

Q: You said last month that you're now looking to focus on two core areas: converting Unix customers to Linux and putting Linux into embedded devices.
A: This is consistent with this shift in focus away from being a visionary company building a great brand with a great concept -- which was that for the first time the customer was given control over the technology that you're asking them to invest in.

Historically the customer ends up being controlled by the vendor because of the proprietary, binary-only nature of the way the industry has been built. When I started (Red Hat) in 1994, this was not enterprise-quality software we were selling. It was highly reliable software, don't get me wrong. But as an example of how far we've come, one of our engineers out of Hungary is leading a project, and built a Web server integrated into the kernel to the point where according to labs tests, this was the fastest Web server technology they had ever seen -- not just the fastest capability for PCs or two-way PCs or four-way, but the fastest they had ever seen. This was rocket science. This innovation came out at the same time that Bill Gates was trying to claim that open source was not an innovative model, it was a copying model.

Are big businesses really going to be interested in switching from Linux to Unix?
What it's allowing us to do is to go into corporate accounts like Morgan Stanley in New York or Toyota over here (in the UK), and offer them not just a lower cost solution but a higher performance and a more reliable solution. The reason we can focus a business vision on Unix to Linux migration is that it is an easy sell.

Our deal with the Institute of Virology and Immunoprophylaxis (IVI) was a great example because it illustrates exactly the point when we're talking about high reliability. They were replacing -- let's just call them legacy PC operating systems on their PCs -- because they work in a clean room. If their machine goes down, they have to exit themselves and their machine out of the clean room, then they have to fix the machine, then they have to go through the whole process of decontaminating themselves and the machine to get it back into the clean room again.

They need reliability. They don't need cost. So they did not go from whatever they were on to a Sun Unix machine, to an IBM AIX machine, they went to a Linux machine because they could build a more reliable system.

The beauty of open source software is that when you run into a bug you can get it fixed. You get source code and you get a license that allows you to modify it. It's like buying a car with a hood that you can open, as opposed to the traditional model in the software industry where the hood is locked shut. If you can open the hood it means you can fix your car, but it also means you have access to 10,000 car repair shops across the UK. Whereas if the hood's locked shut, and if your vendor denies that it's a bug, which is known to happen, you're completely stuck.

If you're IVI and your machine goes down in your lab, you call Microsoft and you say "There's a bug in here somewhere", Microsoft says, "No there isn't, you've done something wrong." And so your only solution is to take the machine out, decontaminate it, and then do the reverse process, and it's a huge pain.

BP (the petroleum company) is putting 3,000 Linux servers at gas stations. What are the odds that if that machine goes down in a gas station there's going to be somebody in 50 miles that can get it working again? And the answer is there just isn't one. Reliability is everything to these guys.

Every Linux company is talking about the enterprise nowadays. What is Red Hat doing differently to compete? Also, do you see any competition coming from IBM's interest in Linux?
In general the thing you have to remember is there's no such thing as The Linux Operating System. Linux is the 16MB kernel of the 800MB operating system that teams like Red Hat engineer. That 800MB consists of -- the last time I looked at it -- 763 different packages that are built by a huge number of different engineering teams across the Internet.

The model ends up looking a lot like a car, in that you don't drive a car, you drive a BMW 525 or a Jaguar X-Type. The point is, you rely on a car company to take all the pieces out of the very creative automotive industry and turn it into a useful product for you. That's what Red Hat does.

IBM has indicated they have no interest in doing that. They understand that the moment they build an IBM Linux operating system, they're going to end up in a bit of a niche, because all the other hardware guys are going to avoid the IBM version in the same way they avoided OS/2.

What about other Linux vendors?
I don't want to denigrate their engineering capabilities, because I know there are users out there who prefer Mandrake or SuSE.

The real problem for the enterprise, though, is that enterprise buyers are not buying tools for today. The job of the Information Services director is not to buy the best solution for the company today; it is to buy the solution that will give the company a five-year return on investment, because he knows he's going to be using that technology for the next five years. If he buys technology from a company that isn't around six months from now, it isn't just his resume that gets hurt, the corporation he's responsible for building gets hurt.

There's no enterprise that's going to rely on the open source community, no matter how smart we all are -- the people contributing to these technologies are all high-level engineers across the Internet. It's going to require a safe solution. This is the benefit of our IPO two years ago, and having put $300m in the bank, which is still there. When we go into a company like Morgan Stanley, we talk to them about having $300m in the bank, having $80m of annual revenue, a rapidly growing and a market cap of $1.5bn, and they say, "these guys are going to be around in five years."

Some of the smaller Linux companies never got traction because they're still perceived by the Morgan Stanleys as being closer to Mark Ewing and I in our spare bedrooms than a company they can rely on, like IBM.

IBM's actually very important to our strategy going forward, because we don't have the man in the van in Lyon, France. So when we win a client in France, and they want somebody to come out and kick his machine for him, that's where having IBM on our side, or having the major system integrators, whether it's HP or Compaq or Dell on our side, is essential for our strategy. If the customer has to go out and buy Microsoft machines from Dell, and rip the Microsoft OS off it and put Red Hat on it, and the only support he's going to get is from a Red Hat employee in Guildford, that's a hard sell.

Caldera, with its acquisition of SCO's Unix business, is now dealing with big enterprises but isn't doing so well. Maybe that says something about where the enterprise Unix market is going.
Caldera made a massive bet. I know Ransom and Brian Sparks over there, I'm sympathetic to the bet they made, but basically they turned themselves from being a Linux company to being a Unix company. They went from doing three or four million dollars in the Linux business to doing $100m of business, 90 percent of which was Unix, which is a proprietary, binary-only operating system.

While 10 years ago it was a state-of-the-art operating system, today, when we talk about a Unix to Linux migration, we aren't really talking about SCO, because SCO is a $100m of a multiple-billion-dollar market, but there's no question that it's one of those opportunities for us.

What makes Linux the future, and not an improved version of Unix, or something like Windows?
In the big picture, it's this collaborative development model.

In 1996 or 1997, we won this big award from InfoWorld, we tied with Windows NT as the best server operating system. The people who were most shocked by this were those of us at Red Hat, because there were 23 of us, including the receptionist, in the tobacco fields of North Carolina at the time. And Microsoft had put a billion dollars into NT, and they'd started the project three years before Red Hat was even formed, and the best they could do was to tie us for the damn award. Who rewrote the laws of economics?

What it hammered home for us was the key value was not what Red Hat was doing, it was this collaborative model we were enabling.

Fast forward from 1997 and where is the community today? You read Slashdot, and people say, "this is all sold out, I'm not contributing my code anymore." We may have alienated the early free software foundation, Richard Stallman's group -- some of those guys are true idealogues, and they contributed software for truly altruistic reasons. But those were always a smaller part of the developer community.

Most people who were contributing software did so in a form of barter system. They needed a better Linux themselves, and that's why they contributed. Don Becker at NASA describes this as clearly as anyone else when he was asked why he contributes extremely fast Ethernet drivers, which is an extremely sophisticated technology, to the Linux kernel, and then allows Red Hat to make money selling his Ethernet drivers, and he doesn't make any money at it. He said, "Let me get this straight: I write a small Ethernet driver, that I admittedly give away, and Red Hat get to put in a box. And in return I get the complete source code and a license to do whatever I want with a complete 800MB operating system, and you're telling me Red Hat's taking advantage of me?"

So it ends up being a true barter system. Well, for every basement programmer we might have alienated, we've gained ten programmers out of SAP or IBM.

That's the power of this model. We literally have the smartest people across the planet working on this stuff. And we have an engineering team that is bigger than anything even Microsoft can put together, much less anything some of the smaller Unix vendors can put together.

The financial environment was pretty terrible last year, but Red Hat didn't do badly. Do you see this improving in 2002, and does it need to improve for Red Hat to keep growing?
In terms of the stock market's view of us, there has been a huge disconnect. The stock market fell in love with Linux stocks for a brief period of time, and has fallen out of love, but meanwhile the Linux market share has continued to grow on a perfectly unabated level. You've watched the IDC numbers over the last five years, you don't notice any form of hiccup.

Some people have always thought the real success of Linux would be if it could challenge Microsoft on the desktop. But that doesn't seem to be Red Hat's focus.
I used to stand up in front of Linux crowds and say, "Linux will never be successful on the desktop," and of course I'd get booed off the stage. And I finally realized the mistake I was making. Linux will not be successful on the PC replacing Windows OS. But we absolutely will be successful on the desktop as a geographic location.

Think about what's the killer app that's taken us as far as we've gone -- because people don't buy operating systems, they buy applications. Well, the killer app that has driven this model is the Internet itself. When you get a collaborative application -- which is the Internet where that collaboration is essential to the value of the network -- then in order for all the players on that network to play fair with each other it has to be open source technology. The moment one company owns a protocol on the Internet, the Internet will fail. It'll be all over. The bulk of the value will disappear.

You could argue that's the idea behind Microsoft's .Net.
Exactly. So if you think about the desktop, if you think about it as a geographic location, how did Microsoft get to own the desktop? Microsoft did not convince people to unplug VMS from their Digital VAX systems in 1979. They took advantage of a major shift in technology toward the PC, and they became the de facto standard on the new technology model, being the PC.

So our opportunity is not to replace Microsoft on the PC. If you've got a perfectly good working PC, why you would go through the angst of replacing it? This is what's kept Apple alive. People like their Apple OS running on their computer. There may be a lot of pressure, there may be a lot of applications on the Windows PC that they don't have on theirs, but theirs works, so why bother?

Editorial standards