In the past 11 chapters, we have covered a lot of ground and taken a look at many concepts and examples relating to operating systems. But studying existing operating systems is different from designing a new one. In this chapter, we will take a quick look at some of the issues and trade-offs that operating systems designers have to consider when designing and implementing a new system.
There is a certain amount of folklore about what is good and what is bad floating around in the operating systems community, but surprisingly little has been written down. Probably the most important book is Fred Brooks’ classic The Mythical Man Month in which he relates his experiences in designing and implementing IBM’s OS/360. The 20th anniversary edition revises some of that material and adds four new chapters (Brooks, 1995).
Three classic papers on operating system design are ‘‘Hints for Computer System Design’’ (Lampson, 1984), ‘‘On Building Systems That Will Fail’’ (Corbato´, 1991), and ‘‘End-to-End Arguments in System Design’’ (Saltzer et al., 1984). Like Brooks’ book, all three papers have survived the years extremely well; most of their insights are still as valid now as when they were first published.
This chapter draws upon these sources as well as on personal experience as designer or codesigner of two operating systems: Amoeba (Tanenbaum et al., 1990) and MINIX (Tanenbaum and Woodhull, 2006). Since no consensus exists among operating system designers about the best way to design an operating system, this chapter will thus be more personal, speculative, and undoubtedly more controversial than the previous ones.