No abstract

In high-performance networks with both advance and immediate reservations, activating an advance reservation may result in the preemption of some ongoing data transfers based on immediate reservations due to insufficient bandwidth. To minimize such service disruption, we explore the interactions between advance and immediate reservations, and propose an integrated reservation solution with two components: (i) a scheduling algorithm based on statistical analysis of reservation dynamics that routes deadline-constrained advance reservations to minimize the probabilistic number of immediate reservation preemptions, (ii) a runtime preemption scheme to minimize the actual number of immediate reservation preemptions at the activation of an advance reservation. Simulation results show that the proposed reservation solution exhibits a superior performance over existing methods.

No abstract

No abstract

No abstract