Comparing a Traditional and a Multi-Agent Load-Balancing System

keywords: Multi-agent systems, distributed systems, load-balancing
This article presents a comparison between agent and non-agent based approaches to building network-load-balancing systems. In particular, two large software systems are compared, one traditional and the other agent-based, both performing the same load balancing functions. Due to the two different architectures, several differences emerge. The differences are analyzed theoretically and practically in terms of design, scalability and fault-tolerance. The advantages and disadvantages of both approaches are presented by combining an analysis of the system and gathering the experience of designers, developers and users. Traditionally, designers specify rigid software structure, while for multi-agent systems the emphasis is on specifying the different tasks and roles, as well as the interconnections between the agents that cooperate autonomously and simultaneously. The major advantages of the multi-agent approach are the introduced abstract design layers and, as a consequence, the more comprehendible top-level design, the increased redundancy, and the improved fault tolerance. The major improvement in performance due to the agent architecture is observed in the case of one or more failed computers. Although the agent-oriented design might not be a silver bullet for building large distributed systems, our analysis and application confirm that it does have a number of advantages over non-agent approaches.
reference: Vol. 25, 2006, No. 1, pp. 17–42