Modification of Traditional Bully Algorithm using Priority Queuing Technique Applied in CPU Memory Allocation

Abstract

Purpose – This study aims to modify the Bully Algorithm, a leader-election algorithm, by introducing Priority Queuing to optimize its steps, and evaluate its efficiency based on message count, election time, and instances of communicating with inactive nodes.

Method – Priority Queuing will organize active nodes in descending order based on their active status, with the election message sent only to the highest-ranked node in the queue. The study intends to compare the performances of three variations of the Bully Algorithm (the Traditional Bully Algorithm, the latest enhancement, and the proposed modification) using a simulator that ensures the algorithms share the same data set.

Result – The findings show that the proposed modification trumps the latest enhancement only during an increased presence of inactive nodes in the distributed system. In return, the newest enhancement trumps the proposed modification when there is little to no presence of inactive nodes.

Conclusion – The proposed modification has successfully reduced the time consumed, communication costs, and the instance of data transmission with failed nodes compared to the traditional method. However, it is not completely better or worse than the latest enhancement.

Recommendation – While conducting the findings for the study, the researchers recommend looking into achieving the same objectives while also considering the reactivation of nodes during the election process. The researchers also recommend fine-tuning the timeout interval and exploring other strategies for enabling multiple nodes to initiate the election process.

Research Implications – This improved algorithm can efficiently coordinate resource management in cloud computing environments, facilitate data replication, and coordinate consensus mechanisms in blockchain networks. These enhancements optimize coordination, fault tolerance, and scalability in distributed systems, ultimately improving performance and user experience.

Practical Implication – The findings of this study have several implications, one of which is enhanced failure tolerance in distributed systems. Moreover, the waiting time-based bully algorithm is an attractive solution for modern distributed systems due to its ability to quickly adapt to network dynamic changes without significant performance degradation.