一种分布式计算中的容错选举算法  

A New Fault-Tolerant Election Algorithm in Distributed Computation

作　　者：赵致琢[1] 黄小炜[1] 吴文鑫[1] 

机构地区：[1]厦门大学计算机科学系,厦门361005

出　　处：《计算机研究与发展》2008年第z1期93-98,共6页Journal of Computer Research and Development

基　　金：国家自然科学基金项目(69383004);福建省自然科学基金项目(A0310007)

摘　　要：选举算法被广泛应用于分布式计算中,而且它已经被证明比合意问题更难.在分析了选举问题和合意问题的关系之后,提出了一种新的容错选举算法.该算法是稳定的、通信有效的,并且该算法可以容忍进程和链路的暂时性错误.所提出的算法不仅解决了选举问题,并且也提供了解决合意问题的一条新的途径.election algorithms are widely used in distributed computing, and it has been proved to be harder than the consensus problem. In this paper, the relation between the election problem and the consensus problem is analysed, and a new fault-tolerant election algorithm is given. This algorithm is stable, communication-efficient, and it can tolerate the temporal failure on the processors and links. The algorithms proposed not only solve the election problem, and it also provides a new way to solve the consensus problem.

关 键 词：选举算法 容错 分布式算法 分布式计算 

分 类 号：TP316.4[自动化与计算机技术—计算机软件与理论]