C

Operating System Concepts

Format: Hardcover
Author: Abraham Silberschatz
ReleaseDate: 14 December, 2004
Publisher: John Wiley & Sons
Rating:

Try reading 3 pages without falling asleep


Sadly, Silverschatz does none of this; in fact, often his book reads more like a tome on tax-law. While "Operating Systems" is not exactly the sexiest subject in Computer Science, it ought to be possible to make it interesting, for example by taking a historical or problem solving approach. Take this sentence, for example:

"If no process is executing in its critical section and some processes wish to enter their critical sections, then only those processes that are not executing in their remainder sections can participate in the decision on which will enter its critical section next, and this selection cannot be postponed indefinitely. " (p. 194, 7th ed. )

Silberschatz also has a tendency to make sweeping statements without giving examples, like what I am doing here. Admittedly, online chapters for different operating systems are available, but I think more examples within the main text itself would have helped to explain the concepts better.

The book also contains errors. For example it says that, "For instance, suppose that the queue usually has just one outstanding request. Then, all scheduling algorithms behave the same, because they have only one choice for where to move the disk head: They all behave like FCFS scheduling. " (p. 461, 7th ed. ) While this is true for shortest-seek-time-first, LOOK and C-LOOK algorithms, it is wrong for SCAN and C-SCAN. They would continue moving the HD head from cylinder 0 to cylinder max, with worse performance than SSTF.

Since I do not have wide experience with other O/S books, I will not give a categorically "don't buy it!" recommendation. After all, Silberschatz is quite comprehensive and could be okay as a reference book. However, if you require a book to teach you O/S concepts, I would strongly recommend looking elsewhere. Perhaps try a book by Tanenbaum? His prose is more readable.

its a text book
Its a lot better than stallings book though. Its a text book a book you have to buy too pass class. For most topics its pretty good explaining, with decent examples. And its a lot better than stallings OS book.

Competent basic text
It covers all the basics: process management, memory, file systems, IO, and communications. This is an adequate text for an introductory operating systems course. It even acknowledges the increasingly important embedded world, and offers case studies of Mach, BSD Unix, and Windows 2000 in on-line support.

That's actually a real strength of the book - its supporting web site. There are appendices, a term's worth of powerpoint slides, assignments, and more. It's very comforting, if you're an instructor without a given syllabus, to have that kind of backup.

The book bottoms out fast, though. It's very weak on processor hardware and how that affects OS design. For example, multithreaded processors, multicores, multiprocessors, and loosely coupled systems have fundamentally different kinds of synchronization issues, which are barely mentioned. Even something as fundamental as "memory mapped IO" gets little, if any mention. This shallowness pervades discussions of networking, file systems with striping and shadowing, and just about everything else. Security may be the weakest section here, except maybe reliability - I'm not sure that's mentioned at all.

But, no matter how fast you talk, there's only so much you can cram into a one-term intro OS course. This is about right, whether you follow the more or less aggressive of the suggested syllabi. One extra star for all the instructor support at the web site - if you need it, you'll appreciate it.

//wiredweird.