An Enhanced Bully Algorithm for Electing a Coordinator in Distributed Systems

Abstract

In a distributed system for accomplishing a large complex task, the task is divided into subtask and distributed among processes and coordination among processes done via message passing. To make proper coordination and functioning we need a leader node or coordinator node which acts as a centralized control node. Leader election is the most challenging task in distributed system because it is not necessary that leader node is always same because of crash failure or out of service may occur in the system. Tremendous algorithms have been proposed for elect the new leader. These algorithms use a different technique to elect a leader in distributed system. Bully election algorithm is one of the traditional algorithms for electing a leader, in which the highest node Id is elected as a leader but this algorithm requires lots of message passing for electing a leader that imposes heavy network traffic. Due to heavy network traffic, it creates complexity in message passing and takes more time. In this paper, we introduce a new approach which overcomes the drawback of existing Bully election algorithm. Our proposed algorithm is an enhanced version of Bully election algorithm. Our analytical result shows that our algorithm is more efficient than original Bully Algorithm.