Queuing Networks with Random Selection for Service

To the class of queuing networks analyzable by the method of Baskett, Chandy, Muntz, and Palacios, we add service centers whose scheduling is random. That is, upon completion of a service interval, the server chooses next to serve one of the waiting customers selected at random. As in the case of first-come first-served (FCFS) scheduling, all tasks must have the same exponentially distributed service time at such a center. We show that for purposes of this analysis, the results are identical to FCFS queuing. Example applications for random selection scheduling in computer system modeling are provided.