SDN-based bandwidth scheduling for prioritized data transfer between data centers

Document Type

Article

Publication Date

8-1-2022

Abstract

The widespread deployment of scientific applications and business services of various types on clouds requires the transfer of big data with different priorities between geographically distributed cloud-based data centers. As a result, Cloud Service Providers (CSP) face a significant challenge to fully utilize the expensive bandwidth resources of the links connecting data centers while guaranteeing Quality of Experience (QoE) for users. Modern data centers are increasingly adopting Software-Defined Networking (SDN) technology, which provides the capability of advance bandwidth reservation. This paper focuses on the collaborative scheduling of multiple prioritized user requests, namely, advance bandwidth reservation with a lower priority and immediate bandwidth reservation with a higher priority, to maximize the total user satisfaction. We formulate this co-scheduling problem with preemption as a generic optimization problem, which is shown to be NP-complete. We design a heuristic algorithm to maximize the number of successfully scheduled requests and minimize the number of preempted advance reservation requests, while minimizing the completion time of each request. Extensive results from simulations with randomly generated networks and emulation-based experiments on an SDN testbed show that our scheduling scheme significantly outperforms greedy approaches in terms of user satisfaction degree, a normalized quantification parameter we define to measure users’ QoE.

Identifier

85112856585 (Scopus)

Publication Title

Cluster Computing

External Full Text Location

https://doi.org/10.1007/s10586-021-03364-7

e-ISSN

15737543

ISSN

13867857

First Page

3019

Last Page

3034

Issue

4

Volume

25

Grant

2019218214GXRC018CG019-GXYD18.2

Fund Ref

Northwestern University

This document is currently not available here.

Share

COinS