A novel threshold-based dynamic load balancing algorithm using mobile agent in distributed system

Abstract

As a fundamental problem in distributed system, load balancing is important to achieving scalability, guarantee system fairness and avoid performance degradation. To efficient utilization of all available resources, we need the concept of load balancing to be added in this area. In message-passing based approaches, the nodes have to exchange messages of load information periodically in order to make decisions on load balancing. The message exchanges result in high communication latency and thus deteriorate the performance of the system. In this paper, we propose a novel threshold-based dynamic load balancing approach which uses mobile agent (MA) because MA based approaches have the merits of flexibility, low network traffic, efficiency, less communication latency, as well as highly asynchronous. The effectiveness of proposed load balancing approach is evaluated by extensive simulation under a dataset including 7000 different connected graphs. Our results show a significant improvement in network traffic compared with the traditional message passing based approach. © 2011 Springer-Verlag.