A new agent based load balancing model for improving the grid performance

Article type: Research Article

Authors: Wided, Ali^* | Okba, Kazar

Affiliations: Computer Science Department, University of Biskra, Biskra, Algeria

Correspondence: [*] Corresponding author: Ali Wided, Computer Science Department, University of Biskra, Biskra, Algeria. E-mail: aliwided1984%****␣mgs-16-mgs200326_temp.tex␣Line␣25␣****@gmail.com.

Abstract: In grid computing, if the system load in each of resources is almost equal, this indicates correct resource usage. This paper presents an agent based load balancing model. Instead of balancing the load in the grid by job migration technique or by moving an entire process to underloaded resources. The proposed agent based load balancing model aims to take advantage of the agent’s characteristics to generate an autonomous system. It also has a distinct advantage over other agent-based load balancing models in terms of resource utilization. The performance evaluation shows that the proposed model can enhance the overall performance of grid computing.

Keywords: Grid computing, job migration, load estimation, performance metrics, grid agent, resource utilization, queue length, cpu usage

DOI: 10.3233/MGS-200326

Journal: Multiagent and Grid Systems, vol. 16, no. 2, pp. 153-170, 2020