Feature Story

Written by Robert McMillan

Friday, 15 November 2002

 

America’s Cup racer, jet pilot, and CEO of the world’s second-largest proprietary software company. Larry Ellison represents many things, but to the Linux community he is best known as an early ally who shares a common enemy: Microsoft.

Oracle may love Linux, and its recently announced cluster file system may be GPL’d, but Ellison’s company is — first and foremost — all about closed source.

Linux Magazine’s Adam Goodman and Robert McMillan caught up with Ellison right after his LinuxWorld keynote in San Francisco this August.

LINUX MAGAZINE: For years now, you’ve downplayed the impact of open source databases. But we have readers who are using MySQL and Postgres. Aren’t those having some impact on Oracle?

LARRY ELLISON: No. An operating system is dramatically simpler than a database — it’s much, much simpler. An operating system can go down, and it’s OK because you reboot. A database can’t go down. You can’t ever lose your data. It’s the one application that has to work all the time. If it succumbs to open source, a whole lot of other things will go before it.

LM: But there are open source databases that people are using — in the Web tier, for example. They already seem more widely adopted than, say, alternatives to Microsoft Office.

ELLISON: But no one in their right mind would use those things for anything real.

LM: So it’s just a question of scale?

ELLISON: It’s not just scale. There’s no security. [There’s no] scalability, security, or reliability. If you give up any of those, you can’t do this. Another reason you’d be out of your mind to use them is because you’d have to buy twice as much hardware. We run, I don’t know, ten times, a hundred times faster than they do on the same hardware.

LM: But don’t you think it’s a question of the specific application of the database?

ELLISON: If this is being used for something that’s tiny — to keep track of your recipes.

LM: But more realistically, do you think that Oracle’s had to cede the lower end of the database market to MySQL to concentrate on the enterprise, or is that a customer base you wouldn’t reach anyway?

ELLISON: No. It’s a little bit like asking if we are worried about people who pirate Oracle. The people who pirate Oracle are not the people who would buy our database. The people who use MySQL are not the people who would buy our database. They don’t have any money. If you have a real application, the first thing you tend to spend money on is a database.

The Linux Threat

LINUX MAGAZINE: When we interviewed him recently, Scott McNealy told Linux Magazine that, when all is said and done, Microsoft may be the company that benefits most from open source software because open source software cuts the profit margins of proprietary software vendors and makes it difficult for them to compete with Microsoft. Do you see things that way?

ELLISON: I think Microsoft’s biggest advantage has been that they’ve been the low-price provider. A lot of people won’t admit that. Microsoft has never had better software, but they’ve had better prices.

And they’ve got huge market share because they run on a low-price platform — Intel — and they’re software is very economically priced. Comparatively crummy, but cheap.

So, I don’t buy Scott’s analysis.

Yes, it’s certainly true that if people are giving stuff away, you can’t spend a fortune on full-page ads in the Wall Street Journal talking about your application server. But I think there’s no better marketing campaign than a high-quality product.

J2EE people aren’t stupid. We’re selling technology to programmers. You can download the stuff and try it yourself. They don’t read the Wall Street Journal anyway. I don’t really buy it.

I think the best way to beat Microsoft is to have better products that cost less. Right now, we have better products that cost more.

LM: Do you think that Sun is having a harder time with Linux than Microsoft?

ELLISON: No. Microsoft has a monopoly. Microsoft can mask any problems that they have because they make so much money. But there’s no doubt that Microsoft has lost more from Linux than anybody else. All of these embedded systems people who are using Linux, would have gone to Microsoft or would have gone to one of the small embedded OS systems.

LM: Microsoft will tell you that it’s easier to migrate from a Unix like Solaris to Linux.

ELLISON: Linux is crucial for all the Unixes. You want people in universities to learn Unix, not Windows. It’s great that you can get a free version of Linux, a free version of Postgres, a free version of this and a free version of that, and that everybody in [academia] can go and play with this stuff and not break any of the laws or steal anybody’s software. And that’s good news, because people tend to like what they know. Otherwise, they might be using Windows.

So I think [Linux] makes the Unix community larger. From our point of view, IBM has so thoroughly blurred what Linux is by running Linux under AIX — you can’t tell what’s really going on there — that it’s just made the Unix community larger.

LM: Are you concerned about IBM’s relationship to the Linux community? They seem to be setting the course of where Linux is going.

ELLISON: No, I think it’s great that IBM’s a big supporter. We’re a big supporter. Dell’s a big supporter. Intel’s a big supporter. Sun is the only one who’s not. There are a lot of big names behind Linux. I don’t think any one company is going to be able to dominate it. In fact, IBM doesn’t want to dominate it. If IBM dominated Linux, and [Linux] lost its cachet of being a cross-industry standard, Linux would dramatically decrease in value. So, IBM had to make a tough decision on that. I know that IBM was weighing whether they should use Red Hat or their own version, and they decided in the end to do Red Hat. That was a very touch-and-go decision inside of IBM because they said, “We could get a competitive advantage by having our own version.” But they backed away from that. It’s way too early for them to try to take control of Linux. You’d want Linux to get a whole lot stronger before you’d try to subvert it.

LM: But would Linux be subvertable at that point?

ELLISON: The stronger it gets, the less likely it is to be subverted. IBM is a strong company, but I don’t think they’re stronger than Intel. In software, I don’t think they’re any stronger than us. So I don’t think any one company is going to be able to grab this any more than any one company can grab the Internet. Plus, the Linux community is so sensitive to the idea of one company taking over, that it just would be antithetical to the community.

LM: I just think that if three years ago, if you said that SourceForge was going to be running on DB2, people would have said, “That’ll never happen.” And now the most popular site for hosting open source projects is running on IBM’s proprietary database. So, the Linux community may be more flexible than people thought.

 

ELLISON: I don’t know if it’s a big deal that the Linux community is getting stuff out of DB2. I suspect that the Linux community, every once in awhile, wanders onto a Web site that’s run by Windows. It’s hard to even know, unless they put [up] a warning. So, that really doesn’t bother me at all. I’m not a big DB2 fan, but I don’t care.

LM: IBM says that it’s important to maintain a diversity of Linux distributions so they can service different regions of the globe. If they want to sell systems in Germany, they sell SuSE; if they want to sell systems in France, they sell Mandrake; in China, it’s Red Flag. Oracle chose to support only Red Hat. Why?

ELLISON: We’ll eventually move to SuSE and Red Flag and Turbolinux — we’re part of the Turbolinux community in Japan. But we have to get one [distribution] up and running and right, so that we can look someone in the face and say, “Look, if anything happens with this system, we’ll fix it. If Red Hat can’t; we’ll fix it. We’ll take full responsibility for every line of code in this thing. We’ll fix it and we’ll submit the bug fix to Red Hat, and it will find its way back into the community.”

The requirements on us are so different than on someone who only has a distribution. [Our customers] say, “We have a problem. You have an hour to fix it.” It’s a different world. So we want to have one distribution that we know absolutely cold — that we’ve thoroughly tested.

LM: Do you talk to Scott McNealy?

ELLISON: All the time. We’re very close friends.

LM: And when you say to him, “Scott, we really think you should be using Red Hat.” What does he say in response?

ELLISON: I talked to him in the last couple of days, and I said, “We should be on the same version of Linux.” He agreed.

We think that Dell, Sun, and Oracle, at the very least, should be on the same version of Linux. And if IBM wants to be there too, that would be even better. We can do a better job of testing. Because in one sense, diversity is a great thing. But the trouble with diversity — the reason every Sony TV isn’t a little bit different — is that it’s very hard to have both diversity and reliability. So it’s great during the development stages: you have all sorts of creative ideas, and that’s wonderful. But at some point, you have to say, “OK, let’s pick one of these things and make a lot of them. This is the Sony TV we’re going to build,” and build millions. So that’s what we’ve done with Red Hat. Pick one, make it work. And it’s possible we’ll go back to SuSE. We’re not locked into any distribution. So we’ll always run on what we think is the best, most reliable, most popular distribution.

LM: So how significant is the fear of Linux forking?

ELLISON: Well, we’ll do everything we can to not let that happen, but that’s part of the strength of the Linux community. I’d be very careful here about giving up what I think you like about the Linux community — which is you have lots of different creative people doing different things. That’s good.

The bad side is, it’s not like the Sony television. So you have plusses and minuses. Our view for all of this is that at any one point in time, we’re going to pick one Linux and say, “we’re going to support this under our database, and we’ll take full responsibility for making this thing work.” In a sense, we don’t think we should even have to tell you which Linux it is. It might be Red Hat this week, it might be SuSE next week, who knows? You shouldn’t care. But we’ll guarantee it works, and it works all the time.

I think you do want to maintain the diversity and the creativity and the distributed development that you currently have, which means lots of different versions, but eventually you’d like to see this stuff kind of come back together. Let a thousand flowers bloom, and then pick one really good bush.

LM: In all likelihood, will that be one distribution?

ELLISON: With Unix, at one time, we had more than a dozen different Unixes back in the days when Unix went all over the place. So I certainly can imagine we could support three different Linuxes or four different Linuxes, but at a certain point it gets ridiculous.

LM: Do you see Linux replacing the proprietary RISC boxes? It seems to have moved up a lot faster than people had expected.

ELLISON: Well, yeah, but I think the really big thing — and this may sound self-serving — [is that] clustering becomes generally accepted. The idea is people would say, “We’d never buy a four-processor box; we’d buy two two-processor boxes, and use a clustered filesystem, which is free.” This is how we operate — we give you better performance. Two-processor boxes are much cheaper than four-processor boxes; way less than half the cost. And two two-processor boxes are much faster than one four-processor box, and these clusters have no single point of failure. They’re much more reliable.

It will really change the landscape of the computer hardware industry, and we’ve bet everything on it. That’s the way we think we win. Because Microsoft and IBM’s databases can’t really do that. They’re shared-nothing; they went down the wrong road. They went down the shared-nothing clustering road, which doesn’t work for real applications. It works for benchmarks. You could write a testing application for shared-nothing, but it’s very hard to write an application for shared-nothing. It’s all custom code.

LM: So is there a better operating system than Linux for that kind of computing?

ELLISON: No. It’s great. Linux is absolutely perfect for that.

Web Services Hype

LM: What do you think of Web services?

ELLISON: We’re the largest provider of technology for building Web services in the world. Web services are the latest colossal hype. People think, “OK, now that I’ve got Web services, I can connect Siebel to PeopleSoft to SAP. We can connect all of these things.” It’s absurd.

What Web services are is a standard protocol that program A can use to talk to program B, if they both agree to speak the same language. People who aren’t programmers just don’t understand why Siebel and SAP can’t talk via Web services. It’s not that they can’t talk the Web services protocol, it’s just what Siebel means by customer is different from what SAP means by customer. SAP is an accounting system that bills customers. It keeps track of your credit rating, how much you owe me. Siebel doesn’t bill customers, it markets to customers. They don’t keep track of that stuff. They don’t have the same information. They can’t communicate.

It’s a little bit as if I said, “I called this guy in France yesterday, and the phone rang. He picked up the phone. We started talking. I had no idea what the hell he was saying. But it made the connection.” And this guy says, “but you’re not using his latest technology. You should have called him on your Web services phone.” It doesn’t solve the problem. And people don’t understand it. It’s actually incredible. People are writing these articles as if this problem is solved, and they clearly have no idea what the problem even is. They’re just clueless. And I’ve seen this happen lots. This is a particularly fun one. I mean, when will this stop?

And I’m a big fan of Web services. I think it’s a very important protocol. All of our applications are Web-service-enabled. We use it constantly. It doesn’t solve all known problems. It’s ridiculous.

Embracing the GPL

LM: Was the decision to open source Oracle’s clustering file system philosophical? Or was it practical?

ELLISON: I think [it was] a little philosophical — that if you want to be an accepted member of the community you have to behave reasonably well. Secondly, it was a new strategy. We want people to use this technology, and they can make it better, and that’s what we expect. That’s what it is to work with the Open Source community. They can find ways to make it better, and the more people who are trying to make it better, the better.

LM: Was there much debate over whether or not to use the GPL license? After all, Microsoft has called that license anti-American.

ELLISON: We have no philosophical differences with the GPL. You don’t have to sign the license. You don’t have to use open source. It’s not anti-American.