Deliveries in global supply chains are often made through lengthy shipping routes that are subject to many delays such as border crossings, inspections and so forth.  Consequently, orders frequently crossover, that is, their order of arrival is not the same as the order that they were issued. In this paper we consider the window fill rate of a multiple location inventory system with order crossover. Demand arrives to each location according to a Poisson arrival process and inventory in each location is reviewed periodically. The system’s window fill rate is the probability that a customer arriving to the system is served within her tolerable wait. We consider two stocking optimization problems. The first is the spares allocation problem, which is defined as finding the optimal allocation that maximizes the window fill rate given a budget of spares. The second problem is the optimal budget problem, which is defined as the optimal allocation that minimizes the total budget of spares subject to a required window fill rate level. We show that when resources are scarce it is optimal to cluster spares to only few locations. In contrast, when resources are abundant, then an equitable solution is optimal.