, University of Michigan
Abstract—Andrew S. Tanenbaum describes the motivation, development, and market impact of the MINIX operating system. The first Web extra at http://youtu.be/86_BkFsb4eI is a video interview in which author Charles Severance speaks with Andrew S. Tanenbaum about the motivation, development, and market impact of the MINIX operating system. The second Web extra at http://youtu.be/U0u4AOx6dd4 is an audio recording of author Charles Severance reading his Computing Conversations column, in which he discusses his interview with Andrew S. Tanenbaum about the motivation, development, and market impact of the MINIX operating system.
Keywords—Andrew S. Tanenbaum; MINIX; Unix; history of computing; Computing Conversations
The story of the evolution of Unix and Unix-like operating systems is very convoluted. Modern Unix-like OSes generally derive from AT&T/BSD Unix or some variation of Linux. But in the beginning, there was only AT&T Unix, and if you trace the moment where the Linux-like OSes flickered to life, you'll find the MINIX OS.
MINIX's original purpose was simply to support university courses that attempted to teach OSes. Essentially, it was a set of floppy disks that came with Andrew Tanenbaum's popular book, Operating Systems: Design and Implementation, first published in 1987. I visited him at Vrije Universiteit (VU), and we talked about the motivation, development, and impact of that simple decision to write an OS to teach a course. Visit computer.org/computingconversations to watch our discussion.
AT&T initially provided the source code to its Unix OS to universities. It didn't have an open source license, but it was readily available, which made it a natural choice when teaching courses about OSes. At first, Tanenbaum used a booklet written by John Lions (John Lions' Commentary on Unix 6th ed., with Source Code; Peer-to-Peer Communications, 1976) to teach his classes:
As an accomplished computer scientist who had been teaching OSes for many years, Tanenbaum knew that writing a simple Unix-like OS wouldn't be all that difficult:
Tanenbaum knew he was close, but he had trouble solving that “one last bug” so he wrote software to emulate the PC hardware:
His OS worked perfectly on the simulator—it could run forever and never crash. Seemingly, it only crashed on the hardware. He was about to give up when he got a lucky break:
Once MINIX was working, Tanenbaum wanted to document it by writing a book similar to the one Lions wrote, something that explained MINIX in great detail so that it could be used to teach operating systems courses:
Even though this was before the Internet, news of the new book spread rapidly. At one point, a bookstore invited him to give a talk that was so well attended it had to use the Santa Clara Convention Center as the venue. A worldwide best-seller, Operating Systems: Design and Implementation, is now in its third edition.
Linus Torvalds was a student who picked up MINIX and started evolving the source code:
In the early days, there seemed to be no reason to view MINIX or Linux as commercial products. Both were academic exercises, while BSD Unix had 20 years of development and production experience. But AT&T wasn't yet done making strategic errors:
While AT&T's lawsuit effectively stalled progress on commercializing BSD Unix, other OSes including Linux began to establish their positions in the market. With all this confusion and with Linux addressing the market for an open source Unix-like system, Tanenbaum kept MINIX's focus on education.
As the field has evolved, Tanenbaum sees that there might be some value in a highly reliable OS:
Because MINIX is based on a very small kernel with most of the OS functionality in user space processes, it's possible to recover from partial failures or even upgrade OS components on the fly without taking the entire system down:
Regardless of MINIX's possible commercial potential, there's no question that when Andrew created MINIX, he set the wheels in motion for an OS market that isn't the exclusive domain of proprietary vendors.
According to chaos theory, sometimes when a butterfly flaps its wings in Brazil it can cause a hurricane in Florida. If Robbert had not casually mentioned interrupt 15, there would have been no MINIX and no Linux and thus no Android (which is based on Linux). The relative stock prices of Apple and Samsung might be quite different now.