No label defined (Q4406375)

This paper deals with the problem of allocation of a Poisson flow of arriving customers to a service facility which consists in \(n\) servers that provide exponential service times on first-in-first-out basis. Classical solutions propose that each arriving customer can examine a group of \(d\) servers in order to join the one of those \(d\) servers having the shortest queue. The ties are broken randomly. This paper provides insight into the fact that breaking ties on the basis of a fixed and asymmetric allocation improves the queueing performance. More concretely, the authors propose the use of fluid limit models and validate their goals with the help of a simulation study. For the entire collection see [Zbl 1003.00011].