Abstract
A distributed system consists of a geographically dispersed collection of computers that are uniquely identified. The problem of providing mutually exclusive access to a critical section in such a system is not trivial. An algorithm is proposed that provides mutual exclusion even in the event of node or communication failure. Also presented is a method that allows a node to reenter the system upon recovery. It is assumed that all nodes communicate by message passing and do not share memory. Because there is no global clock, the messages are ordered by use of time stamps. The algorithm maintains mutual exclusion, is deadlock free, and prevents starvation. Adapting the recovery aspects of the algorithm to other distributed mutual exclusion algorithms and extending it to handle multiple failures are discussed.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
