We explore the issues of when and how to partition arriving customers into
service groups that will be served separately, in a first-come first-served
manner, by multiserver service systems having a provision for waiting, and
how to assign an appropriate number of servers to each group. We assume th
at customers can be classified upon arrival, so that different service grou
ps can have different service-time distributions. We provide methodology fo
r quantifying the tradeoff between economies of scale associated with large
r systems and the benefit of having customers with shorter service times se
parated from other customers with longer service times, as is done in servi
ce systems with express lines. To properly quantify this tradeoff, it is im
portant to characterize service-time distributions beyond their means. In p
articular, it is important to also determine the variance of the service-ti
me distribution of each service group. Assuming Poisson arrival processes,
we then can model the congestion experienced by each server group as an M/G
/s queue with unlimited waiting room. We use previously developed approxima
tions for M/G/s performance measures to quickly evaluate alternative partit
ions.