Torvalds on Linux

Torvalds on Linux

A year or two ago the open-source revolution was a mere glimmer in the eyes of some devoted Linux developers. Today, companies that previously snubbed the freely distributed and freely modified operating system are proudly announcing Linux ports and strategies to open up their own code.

The man behind Linux, 29-year-old Linus Torvalds, broke ranks with his family of journalists in Helsinki to become a programmer early in life. Eight years after releasing Linux, Torvalds is still in the trenches, working at secretive Transmeta Corporation for the past two years. In this interview with Elinor Mills Abreu, Torvalds talks about the open-source revolution and the benefits of different Linux distributions and development partnerships.

IDG: What is Linux's road map for the future?

Torvalds: There's never really been a road map. In the sense that the Linux user base has been changing fairly rapidly, making a five-year plan just would not work. A year ago the main user for this was still on a kind of technical workstation, a small-scale Web server. And suddenly enterprise-like large-scale computing arrived. It wasn't something that Linux had really been used in but it meant that suddenly there was a lot of new user interest in a completely new area. So we're moving on to doing better and better things and it's not really planned. It's more of a reaction to what people need.

How many Linux users are there?

I have no idea. Nobody really counts. There's been a number of guesses, some of them made by doing samples and some of them by asking the vendors, but the user base seems to be around 10 million plus.

What can you tell me about the plans for Linux 2.4 and even for 3.0?

Well, right now the current tree is obviously 2.2. What happened in 2.2 is that we have developed a fairly stable and well-designed SMP (symmetric multiprocessing) subsystem, for example. But it hasn't been really tuned or tweaked to any degree. I think that the goal for 3.0, which is probably going to be the next version, the next stable version in a year-and-a-half, is to have SMP that really scales up to eight or 16 CPUs (central processing units) and probably some clustering support. Right now we're at two and four being a reasonable upper limit and we want to take that to at least eight or 16. This is kind of what I've been working on. There's a lot of people interested in, for example, real-time issues for small controllers, where you want to have an industrial controller or even an embedded system and then you have completely different needs.

I'm really happy that the whole Linux development model kind of depends on the fact that there's a lot of people around and you don't get in a rut. You don't start concentrating on some marketing scheme goal. There is no one goal. There's like 10,000 different people with different ideas and some of them obviously are not very good. but the good ones eventually percolate up to the top.

How will a choice be made among different proposed successors, and by when?

Some people call the Linux model kind of communistic or whatever, but actu-ally it's very aggressively competitive in the marketplace.

It's just that the competition is, to a large degree, about technical merit and basically what happens is that a lot of people are developing a lot of stuff. It's not just the kernel, it's all these other things, too. And that which works wins, basically.

Are you concerned that the focus on Red Hat will affect future adoption of Linux?

No. I'm not concerned. One of the complaints that commercial people in particular had about Linux originally was that there was nobody to point the finger at.

There was no one entity to take the blame or even to take the cheque to buy more licences. To a large degree that is a psychological problem and to a large degree having Red Hat as a big name is good for that part of it. That's why management people like Microsoft. It feels safe to buy from Microsoft.

Which independent software vendor has surprised you the most by announcing a Linux port?

None of the big ones have been that big of a surprise. By the time they announced the jungle drums had been drumming for a while. All the major ones have been making noises in private - not usually to me because I'm not even that interested.

But they have been in discussion with people about it. And in some cases the timing was a surprise. For example, Oracle was very ready to do the port but then I think it was Informix that just forced their hand.

How did Informix force their hand?

Informix was the first major database vendor to announce publicly support for Linux and I think Oracle within a week said, 'Okay, we'll do it too'. Once Oracle announced then everybody else was on the verge of saying something. In some way it was politically OK and everybody felt safe about saying, 'OK, we're also offering Linux support'.

We've read that you think Microsoft will jump on the Linux bandwagon.



I think Microsoft is so essentially market driven and they have ported their applications to the Macintosh, that's despite the fact that . . . the Macintosh had like 10 per cent of the market. Maybe for political reasons they wanted to wait until Linux has 15 per cent of the market.

Hey, it will happen. They'll go where the money is.

Why did you develop Linux?

Because developing it was a lot of fun and that's really how it got started. And I needed something like Linux so the main question was really what should I develop next because I was a programmer and I wanted to have a program to work on.

Do you think the Linux phenomenon will continue?

To some degree. The kind of growth that happened three, four years ago where you had 10-fold increase every year - that will certainly not happen. I think it was you guys who suggested that 25 per cent annual cumulative growth was expected. That's enough for me.

There's a lot of reasons why people want to go to Linux. The technical ones being fairly well known, but there's also a lot of political reasons why. A lot of people just do not like Microsoft and the other reason is purely economic - that it's really hard to compete against Microsoft, unless you do it on your own terms. So obviously a lot of companies see Linux as a way of competing without having to go head-to-head with Microsoft.

Follow Us

Join the newsletter!

Error: Please check your email address.
Show Comments