Browse Books

Reviewer: R. Clayton

In the preface, the authors claim that this book “provides a clear description of the concepts that underlie operating systems” (p. vii). They go on to distinguish this book from its progenitor, the fifth edition of Operating system concepts [1], which “presented concepts fundamental to all operating systems, rather than particular implementations of systems. This text discusses applied knowledge” (p. x). It might be possible to reconcile these two descriptions, but I will just say that the first description is more appropriate than the second. The book contains 22 chapters divided into six parts. Part 1 is a general introduction and an introduction to the structures of computer systems and operating systems. Part 2 presents process management, with chapters on processes, threads, scheduling, synchronization, and deadlocks. Part 3 is dedicated to storage management. It covers memory management, virtual memory, file systems, I/O systems, and mass storage structures. Part 4 covers distributed systems in terms of network structures, distributed communications, distributed coordination, and distributed file systems. Protection and security each get a chapter in Part 5, and Part 6 contains case studies of the Unix, Linux, and Windows NT operating systems. Each chapter ends with exercises and a bibliography. The exercises mostly test recall; there is no answer section. The bibliography covers references of historical and modern interest and of both theoretical and practical interest. A pair of related Web pages contain supplementary material, including extra chapters, course material, and an errata sheet. Applied operating systems knowledge is presented using examples drawn from current commercial operating systems, mainly Solaris, Windows NT and, to a lesser extent, the Macintosh OS. Applied knowledge is also described by code fragments in Java. The operating systems examples are detailed and relevant. They are, however, spread throughout the book, and some rearrangement may be necessary to get the best classroom use out of the text. For example, instructors who want to include the Solaris Virtual File System in a section on file systems will have to detour to the chapter on distributed file systems. Such problems cannot be entirely resolved, because there is no one best order, but they can be minimized by good cross-referencing, of which this book needs more, particularly between the case studies in Part 6 and the rest of the book. The index is good, and a diligent and patient reader will be able to navigate through the book, but casual, hurried, or less knowledgable readers may have difficulty discovering everything the book has to offer. Java appears to have been chosen because of its thread and synchronization facilities; a majority of the Java examples appear in Part 2, on processes management. Java is a fashionable choice, but one that has practical problems. In particular, Java's thread synchronization facilities are semantically weak, and it can be difficult to use them as a basis for implementing stronger synchronization facilities [2]. Writing correct synchronization facilities is a notoriously difficult problem, and it is not a strong mark against Java that it does not do much to simplify the problem. This difficulty suggests that writing correct synchronization facilities—as opposed to correctly using synchronization facilities—is beyond the scope of general-purpose operating systems textbooks. Putting this suggestion aside, this book neither distinguishes nor disgraces itself in this area relative to a majority of its peers. With no compelling positive or negative points to its credit, the value of this book is reduced to personal choice. I find it better organized than the textbook I am currently using [3], but weaker in applied knowledge, because that book contains the source code of a complete operating system. Some people may perceive and object to a Unix bias; two of the three case studies, for example, are Unix-based. The book ranges widely into relevant areas, covering, for example, stable storage and CORBA. This wide range, however, does impose some pruning chores on those whose time is limited, and may make the book unwieldy for self- study. Overall, this book supports my operating systems textbook conjecture: if you own one operating systems textbook, you own them all.