An improved reliable distributed mutual exclusion algorithm

Abstract

Reliability and efficiency are two important characteristics of a good distributed mutual exclusion algorithm. Efficiency is generally considered in terms of the number of message transfers required. Considering the reliability aspect, it is desirable that an algorithm should be 'fully fault tolerant', i.e. capable of providing failed node removal and restart facility. Raymond's algorithm is efficient and it needs only O(log N) messages to enter a critical section. The authors show that Raymond's algorithm is not fully fault-tolerant and improve it to provide fully fault-tolerant capability for the case of single node failure.<>