Improving the Performance of Bully Algorithm for Leader Election in Distributed Systems

Leader election is a classical problem in distributed systems in which the selection a coordinator node to perform tasks is necessary. However, coordinator can fail and this requires performing a leader election operation to choose a new coordinator. This study focuses on the Bully algorithm which uses a lot of messages and higher network traffic which are considered the main drawback. This paper provides a simple and adaptable adaptive bully algorithm that lowers the messages passing used and is simple and flexible. The algorithm utilizes only one node, called responsibility for the election, and a time span, where a time span is set for each node. These two procedures minimize leader election operations especially discovering failures from multiple processes in the system and messages passing. According to the results, the enhanced algorithm is better than the algorithm bully. It reduces the complexity of message passing during leader election operations.