A state-aggregation method for analyzing dynamic load-balancing policies

Exact performance analyses of dynamic load-balancing policies for distributed systems are very difficult because the state space is multidimensional and load-balancing decisions are state-dependent. A state-aggregation method is proposed to analyze the performance of dynamic load-balancing policies. Those states with the same number of jobs are aggregated into a single state. The number of jobs in the system is modeled by a birth-death Markov process. The state transition rates are estimated by an iterative procedure. The proposed state-aggregation method is applied to analyze the performance of a particular dynamic load-balancing policy, namely a symmetric policy with threshold value equal to one. Extensive simulations were performed to study the accuracy of the state-aggregation method. This method provides accurate performance estimates for the symmetric policy for systems of various sizes when the mean job transfer delay is small compared to the average job service time