Master’s Thesis at the College of Information Technology Discusses a Dynamic Approach for Reducing Idle Time and Enhancing Load Balancing with Fault Tolerance in Computing Systems

Duhaa Fadill Abbas
The Department of Software at the College of Information Technology hosted the defense of a master’s thesis entitled “A Dynamic Approach for Reducing Idle Time and Enhancing Load Balancing with Fault Tolerance in Computing Systems.” The thesis was presented by the graduate student Ahmed Abdul-Jabbar Hassoun and supervised by Dr. Mahdi Saleh Naamah. The defense took place on Thursday, May 22, 2025, at 9:00 a.m. in the conference hall of the college.

The thesis explores the field of grid computing, which represents a vital and rapidly evolving domain within computer science, aiming to meet growing demands for performance and reliability in distributed systems. Despite its potential, grid computing still faces critical challenges such as idle resources, inefficient task allocation, and limited fault tolerance capabilities. Traditional load balancing methods are often incapable of adapting to real-time changes in server performance, which hinders the efficiency of such systems.

In response to these challenges, the study proposes a dynamic, data-driven model for fault tolerance that seeks to enhance the efficiency of grid computing environments. The approach integrates mechanisms for analyzing network characteristics, selecting servers based on stability and responsiveness, and redistributing workloads intelligently according to both hardware and network conditions. The core of the model is an adaptive algorithm designed to assess server capabilities and allocate requests accordingly, while also incorporating a fault-tolerant strategy based on shared backups and system checkpoints to ensure service continuity and resilience.

The model’s effectiveness was evaluated through key performance indicators such as idle time reduction, load distribution efficiency, scheduling fairness, and energy consumption. Experimental results revealed that the proposed system significantly outperforms several conventional algorithms, achieving up to a 57.62 percent reduction in idle time compared to methods like Round Robin, Weighted Round Robin, Least Connection, and Consistent Hashing.

This thesis contributes to the development of an intelligent and scalable framework for dynamic load balancing and fault tolerance within grid computing systems, with promising potential for future expansion into cloud and autonomous computing environments.