A performance comparison of adaptive and static load balancing in heterogeneous distributed systems

This paper focuses on using simulation to compare the performances of adaptive and static load balancing policies in a heterogeneous distributed system model. All the hosts (nodes) an the system are assumed to have the same function but possibly different processing capacities. The overheads and the delays for both job transfer and system state-information exchange are assumed to be nonnegligible. Simulation results show that both adaptive and static policies improve performance dramatically, and that the performance provided by static policies is not much inferior to that provided by adaptive policies. They also show that when overheads are nonnegligibly high at heavy system loads, static policies can provide performance more stable and better than that provided by adaptive policies