Intelligent and Flexible Bandwidth Scheduling for Data Transfers in Dedicated High-Performance Networks

Abstract

High-demanding transfers of extremely large amounts of data have been increasingly supported by the bandwidth reservation services in dedicated high-performance networks (HPNs). To use the bandwidth reservation service, a user needs to initialize a bandwidth reservation request (BRR) containing parameters and requirements of the bandwidth reservation or data transfer, such as the time interval. Upon receiving one BRR, existing researches and bandwidth reservation service providers return either the best-case bandwidth reservation (BR) or a reject message if not all the BRR requirements can be satisfied. In this paper, we study the intelligent and flexible bandwidth scheduling of two common types of BRRs: direct bandwidth reservation and indirect bandwidth reservation (data transfer). When not all the requirements of the first type can be satisfied, instead of a direct reject message, we identify and return two alternative BRs made within the closest time intervals before and after the user-specified time interval; for the second type, besides the above two alternative BRs, we also identify and return the alternative BR made within the closest time interval crossing the user-specified time interval. For each problem, we design a flexible bandwidth scheduling algorithm with rigorous optimality proofs to compute both the best-case and alternative BRs. For comparison, we also design two heuristic algorithms adapted from existing bandwidth scheduling algorithms. We then conduct extensive simulations to compare their overall performance, and the simulation results show that the proposed optimal algorithms have superior performance to those in comparison. To the best of our knowledge, our work in this paper is among the first to study intelligent and adaptive bandwidth scheduling with alternative BRs in HPNs.