Andrew Tanenbaum: Operating systems' Mr Reliable

MINIX 3 is quite different to versions 1 and 2. Where MINIX 1 and 2 were intended as teaching tools in universities, MINIX 3 targets resources-limited systems, and also embedded computing. What took you down this path and how has it been received?

It is not only resource limited systems. I am trying to get away from massive bloat even on desktop PCs. A Windows or Linux system takes a minute or two to boot. Version 7 UNIX booted in about 10 seconds on a PDP-11 that was 2000 times slower than a modern PC. Personally, I don't think this is progress. There is simply far too much bloated unreliable code around. There is no way you are ever going to get, say, 3 million lines of kernel code bug free, when experimental studies have shown there to be something like 10 bugs per 1000 lines of code and drives have 3-7x more bugs than normal OS code.

With that in mind, are students still the main users/target audience for the MINIX software today?

MINIX 3 is still pretty young. It has been heavily redesigned in the past two years. The goal is a small, acceptably fast, highly reliable system that can repair itself on the fly and hopefully never crash. The target could be embedded systems, but also things like PCs and smaller devices. But as I said, we have only been at it for about two years with a fairly small crew, so we are far from done.

Where do you ultimately want to go with MINIX?

I would like to demonstrate how to build a highly reliable system, and I don't think putting millions of lines of code in the kernel is the way to go.

Almost 20 years on, what would you say has been MINIX's contribution to society?

Well, it has educated hundreds of thousands of students worldwide in operating system design. It also led to Linux. I doubt that Linus would have written Linux if he hadn't had the MINIX source code to study carefully. I think that it will yet show that the only way to make a system truly reliable is to make it small and modular.

Leading from that question, MINIX proved to be an influential stepping stone to the now incredibly popular and widely distributed Linux OS. How does that make you feel?

Professors like it when their students -- be they local or remote -- do well.

You have been very protective of MINIX, which in its early period came under fire for all sorts of reasons -- whether it be the fact you charged for it, or because you have tried to keep it simple as possible. What has this OS personally meant to you?

There is really a vision behind MINIX and I have objected when people have tried to add all kinds of features and make it more bloated. There is much to be said for that. I have run Linux from time to time and every time I tried to install an application it failed due to it not finding some obscure library I never heard of. There are 100+ Linux distributions out there. More centralized control would have avoided this mess. There is only 1 Windows XP due to Microsoft's maintaining tight control. This is one thing they got right.