Data Transfers Using Bandwidth Reservation Through Multiple Disjoint Paths of Dynamic HPNs

Abstract

Bandwidth reservation was initially designed to provide QoS for real-time multimedia applications, and has been extended to many other applications and widely used because of its proved high efficiency and effectiveness. Particularly, it has now become an indispensable service to support big data transfers in dynamic high-performance networks (HPNs) for remote data storage and processing. A number of problems regarding data transfers using bandwidth reservation have been investigated in the past few years, and most of them focused on the data transfers on one path or two disjoint paths of the dynamic HPNs. In this paper, three important problems regarding data transfers using bandwidth reservation on multiple node- and edge-disjoint paths in dynamic HPNs to achieve are studied: (i) the earliest completion time for a data transfer request through multiple disjoint variable paths with variable bandwidths, (ii) the earliest completion time for a data transfer request through multiple disjoint fixed paths with fixed bandwidths, and (iii) the minimum number of disjoint fixed paths with fixed bandwidths for a deadline-constrained data transfer request. We prove all of these problems to be NP-complete and propose one heuristic algorithm for each. We compare the proposed algorithms with two existing scheduling algorithms and one algorithm derived from an existing algorithm, and conduct extensive simulations. The simulation results show that our proposed heuristic algorithms have much better overall scheduling performance.