Recently scientists in Yale University published their research on explaining why computer crashes and we don’t. The compared the Linux operating system and a kind of bacterium called Escherichia coli. The scientists found out both systems are hierarchical but organized in different ways. The organization of a bacterium is like a pyramid with a few overheads controlling many primitives. While in the case of OS, there are few primitive routines but lots of high-level functions invoking them. Hence, any breakdown of such primitive routines would result in a much larger number of high-level crashes. That seems to explain why computer crashes all the time but we don’t.
Well, in my opinion, OS is not a bacterium so it is pointless to compare them. An OS is a platform enabling user to put on their own plays on it. Hence it provides a number of primitive routines so that higher-level user (e.g. software developers, non-technical users) can create fancy functions and applications and perform various kinds of tasks effectively. That’s why we can see that an OS provides only a small number of primitive routines which are used by a much larger number of high-level functions. And that’s the way it should be. If an OS should be like a bacterium with a pyramid-like structure, then we cannot imagine the growth of its size, as software and applications grow in an enormously high speed. One solution might be isolating high-level invokes so that an OS would look like many *bacteriums* functioning together, each with a stable pyramid-like structure. However this is unlikely to happen since software and application are dependent on each other: they compete for CPU, memory and disk storage. That’s why crashes happen. As cloud computing is emerging, can we imagine having these high-level stuff located on different remote servers? And each server is like a *bacterium*, solely concentrating on one single software or application. If this is possible, most crashes might be eliminated.