Chapter-18 - M18_STAL6329_06_SE_C18.QXD 2/22/08 8:45 PM...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon
18-1 D ISTRIBUTED P ROCESS M ANAGEMENT 18.1 Process Migration Motivation Process Migration Mechanisms Negotiation of Migration Eviction Preemptive versus Nonpreemptive Transfers 18.2 Distributed Global States Global States and Distributed Snapshots The Distributed Snapshot Algorithm 18.3 Distributed Mutual Exclusion Distributed Mutual Exclusion Concepts Ordering of Events in a Distributed System Distributed Queue A Token-Passing Approach 18.4 Distributed Deadlock Deadlock in Resource Allocation Deadlock in Message Communication 18.5 Summary 18.6 Recommended Reading 18.7 Key Terms, Review Questions, and Problems CHAPTER M18_STAL6329_06_SE_C18.QXD 2/22/08 8:45 PM Page 18-1
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
18-2 CHAPTER 18 / DISTRIBUTED PROCESS MANAGEMENT This chapter examines key mechanisms used in distributed operating systems. First we look at process migration, which is the movement of an active process from one ma- chine to another. Next, we examine the question of how processes on different systems can coordinate their activities when each is governed by a local clock and when there is a delay in the exchange of information. Finally, we explore two key issues in distributed process management: mutual exclusion and deadlock. 18.1 PROCESS MIGRATION Process migration is the transfer of a sufficient amount of the state of a process from one computer to another for the process to execute on the target machine. Interest in this concept grew out of research into methods of load balancing across multiple networked systems, although the application of the concept now extends beyond that one area. In the past,only a few of the many papers on load distribution were based on true implementations of process migration, which includes the ability to preempt a process on one machine and reactivate it later on another machine. Experience showed that preemptive process migration is possible, although with higher overhead and complex- ity than originally anticipated [ARTS89a]. This cost led some observers to conclude that process migration was not practical. Such assessments have proved too pessimistic. New implementations, including those in commercial products, have fueled a continu- ing interest and new developments in this area.This section provides an overview. Motivation Process migration is desirable in distributed systems for a number of reasons [SMIT88, JUL88], including: Load sharing: By moving processes from heavily loaded to lightly loaded sys- tems, the load can be balanced to improve overall performance. Empirical data suggest that significant performance improvements are possible [LELA86, CABR86]. However, care must be taken in the design of load-balancing algo- rithms. [EAGE86] points out that the more communication necessary for the distributed system to perform the balancing, the worse the performance be- comes.A discussion of this issue, with references to other studies, can be found in [ESKI90].
Background image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 09/13/2008 for the course CSCI 4020u taught by Professor Jimcar during the Spring '08 term at Trinity University.

Page1 / 36

Chapter-18 - M18_STAL6329_06_SE_C18.QXD 2/22/08 8:45 PM...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online