The stated goal of this book is to “develop an efficient virtual memory system model that supports a quantitative comparison of local and global memory management policies.”
The initial two chapters provide an introduction, motivation, and review of virtual memory system management methods. In the second chapter, the discussion is divided into three sections: scheduling processor and memory, local and global memory management policies, and load control as an interface between scheduling and memory management.
The final four chapters discuss virtual memory models, simulating virtual memory computer systems, and empirical studies of model preparation and utilization. After an extensive review of program behavior and computer system models, Chapter 3 builds a case for utilizing simulation models with an explicit representation of page referencing and main memory allocation. Chapter 4 describes an instance of such a model for a trace-driven discrete-event simulation of a computer system. Chapter 5 discusses the use of the model in Chapter 4 to study virtual memory management policies. The new algorithm, WSCLOCK, which was introduced in Chapter 2, is shown to have very good performance.
The author does accomplish his stated goal. He also demonstrates that such simulation models can be efficient and useful.
Carr has done an excellent job of constructing the interrelated pieces of this book. For example, the methods discussed in Chapter 2 form the basis for the operating system model discussed in Chapter 4. The first two chapters of the book would be excellent additional material for an operating systems overview course. The entire book could be used as the basis for a course on systems modeling. I strongly recommend this book.