A Population-based Approach for Load Balancing in Distributed Computing Systems

For achieving high performance in distributed computing systems which may consist of a set of servers and a set of self-interested users, it is required to balance the load between servers efficiently. In these systems, each user wants to assign its jobs to servers in a selfish-manner with the aim of minimizing its response time. No user is interested in optimizing social welfare (e.g., the overall response time of the system) and fairness (e.g., equality in the response time). However, while balancing the load among servers, we need to consider these three (i.e., individual expected response time, total system-wide response time, and fairness) fundamental issues of the systems. In view of these issues, we propose a population-based algorithm that undertakes all these issues into consideration and provides an optimal solution to the problem. To assess the efficacy and relative performance of our algorithm, we benchmarked it against existing schemes. Finally, numerical results advocate its supremacy over existing approaches.