Task Scheduling in Parallel and Distributed Systems (Prentice...

Features

• Cover Type: Hard Cover with 304 pages
• Published by: Prentice Hall May 1994
• Written in: English
• ISBN 10 Number: 0130992356
• ISBN 13 Number: 978-0130992352
• Book Dimensions: 9.5 x 6.5 x 1 inches
• Weighs: 1.4 pounds

The publisher, Prentice-Hall ECS Professional
El-Rewini and Lewis were among the first researchers to recognize the problem of resource allocation (scheduling) inherent in parallel and distributed programs -- those which consist of interacting tasks, each capable of running on a separate processor. Without proper allocation and ordering of tasks, the parallel and distributed program will not deliver improvements beyond that of a serial program. Until now, much of the literature on this problem has been hidden in unusual research journals. This single-source reference addresses the resource allocation problem in all of its variations, and surveys the most important scheduling techniques that have been proposed over the past decade.

Reader Reviews
You can summarize the role of any operating system as a queue manager, which makes scheduling the most important component. Scheduling in single-processor or single-system environment, done right, is elegant. Crafting *elegant* scheduling algorithms for multi-processor, parallel-processing or distributed systems is another matter entirely. This book provides the foundation of knowledge and techniques for such an undertaking. What sets this book apart from other highly technical tomes on narrow topics is the fact that it's extremely well organized, is clearly written and has illustrations that closely support the concepts set forth. The algorithms provided epitomize elegance in coding and efficiency. The book is organized into chapters that are aligned to a scheduling taxonomy as follows: Chapter 1 is the root, and discusses the classical scheduling problem and a general model of a scheduling system. Deterministic scheduling in the Multiple Instruction Multiple Data (MIMD) family with precedence relations among tasks is the topics of chapters 2 through 6 and ten through 13. Optimal algorithms are covered on chapters 2, 3 (without communication) and 4 (with communication). Chapters 5, 6 and 10-13 cover heuristics. Chapter 7, again in the deterministic family, is devoted to single program multiple data (SPMD) scheduling, while non deterministic scheduling is covered in chapters 8 (static) and 9 (dynamic), with coverage of hybrid non deterministic algorithms given in chapters 8 and 10. Scheduling with no precedence relations (i.e., allocation) is the topic of chapter 14. This is one of the best books on scheduling that I've read. It goes much deeper than books on operating system design and is surprisingly better written than most I've read, including Comer's and Tanenbaum's popular works. If you want to understand the complexities of parallel scheduling I highly recommend spending the time and effort to track down a used copy of this book and use it as a starting point.