https://www.dropbox.com/s/7shjro7gyqne6fd/Screenshot%202019-04-20%2018.19.48.png?dl=0

Abstract

Load-balancing algorithms play a key role in improving the performance of practical distributed systems that consist of heterogeneous nodes. The performance of any load-balancing algorithms and its convergence-rate is affected by the structural factors of the network that executes the algorithm. The performance deteriorated as the number of system nodes, the network-diameter, the communication-overhead increased. Moreover, additional technical-factors of the algorithm itself significantly affect the performance of rebalancing the load among nodes. Therefore, this paper proposes an approach that improves the performance of load-balancing algorithms by considering the load-balancing technical-factors and the structure of the network executes the algorithm. We applied the proposed method to a cafeteria system in a university campus and compared our approach with two significant methods presented in the literature. Results indicate that our approach considerably outperformed the original neighborhood approach and the nearest neighbor approach in terms of response time, throughput, and communication overhead.