Allocation of distinguishable servers (Q1820674)

The model considered is a variation of the hypercube model in which there are R distinguishable servers and N types of customers. We consider two interesting special cases - light and heavy traffic intensities - and use asymptotic techniques to derive policies for assigning servers to arriving customers which are optimal under these conditions. For light traffic, we show that it is optimal to assign the server with the smallest assignment cost while, for heavy traffic, we derive an efficient algorithm for finding the optimal policy.