Effects of Different Priority Policies on the Capacity Design for Multiclass

We consider an IvI Ilv!1 c queue with two classes of customers. We examine the effect of different priority policies on the optimal number of servers to minimize the long-run expected average server and holding costs of the system. Three different policies are considered; the non-idling non-preemptive and preemptive policies as well as the idling policy which minimizes the long-run expected average holding cost of the system in the class of non-preemptive policies. We show that their effect on the optimal number of servers is not significant for many real life systems. This conclusion allows us to consider the design problem under any of the above policies, which introduces significant computational simplification for the system under the idling policy.