Network Load Balancing with Strong Migration in an Agent Based Grid System Using CSP Approach

In this paper we present a dynamic network traffic balancing approach using strong migration on an agent based grid environment. The paper focuses on three different areas, namely load balancing, strong mobility and CSP (Constraint Satisfaction Problem) approach. We use CSP approach for making task migration decisions. Strong task migration is used to move tasks between nodes at runtime to maintain dynamic balancing. Our approach is implemented on an agent based grid system, where all loads can be formulated as weighted constraints of a CSP. We define inter-messaging rates of tasks as the main load of our system. Activated by the results produced by CSP execution, the strong migration mechanism we have integrated into the grid system maintains dynamic traffic balancing by transferring tasks such that frequently communicating tasks end up on the same node and do not consume network bandwidth. Experimental results show remarkable reduction on network use of the grid system