A Centralized Token-Based Algorithm for Distributed Mutual Exclusion

In this paper, we present a new algorithm for distributed mutual exclusion that combines the advantages of existing centralized and token-based algorithms: it takes only two messages to acquire the lock when it is not held by another process, and only one message to pass the lock from one process to another. At the same time, our algorithm generates signiicantly less message traac than non-centralized token-based algorithms. We also show that, within our approach, the performance of the algorithm can be further improved by making use of information on the pattern of lock requests, if available. Finally, we outline a combined protocol that uses our eecient algorithm most of the time and switches dynamically without service interruption to an existing non-centralized token-based algorithm during periods when the central node in our algorithm becomes temprorarily overloaded.