Minimizing mean response time in non-observable distributed systems with processor sharing nodes

Consider a non-observable distributed processing system with N ≥ 2 single server queues operating in parallel, each under the processor sharing discipline. Jobs arrive one by one to the dispatcher, which immediately routes it to one of the queues. When making a routing decision the dispatcher does not have any online information about the system (current queues’ sizes, size of the arriving job etc.) The only information available to the dispatcher is: job size distribution, job’s inter-arrival time distribution, server’s speeds, time instants of previously arrived jobs and previous routing decisions. Under these conditions, one is interested in the routing policies which minimize the job’s long-run mean response time. Two new class of policies are being proposed, which, according to the numerical experiments, may significantly outperform the optimal probabilistic policy and the Round Robin policy. ©ECMS Mauro Iacono, Francesco Palmieri, Marco Gribaudo, Massimo Ficco (Editors).