King's College London

Research portal

Joint reactive and proactive SDN controller assignment for load balancing

Research output: Chapter in Book/Report/Conference proceedingConference paper

Original languageEnglish
Title of host publication2019 IEEE Globecom Workshops, GC Wkshps 2019 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728109602
DOIs
Publication statusPublished - Dec 2019
Event2019 IEEE Globecom Workshops, GC Wkshps 2019 - Waikoloa, United States
Duration: 9 Dec 201913 Dec 2019

Publication series

Name2019 IEEE Globecom Workshops, GC Wkshps 2019 - Proceedings

Conference

Conference2019 IEEE Globecom Workshops, GC Wkshps 2019
CountryUnited States
CityWaikoloa
Period9/12/201913/12/2019

King's Authors

Abstract

In Software Defined Networks (SDN), the controller is considered as a critical network element with respect to the overall operation of the network. The inherent centralized nature of the SDN controller brings sufficient flexibility to network management, but in the case of congestion episodes or failure, the whole system can be affected. In that respect, the spatiotemporal variation of the network traffic affects the network performance by increasing the response time of the control plane when it is overloaded, raising in that respect the issues of reliability and scalability. In this work, we are aiming to tackle the problem of load balancing in the control plane. The proposed approach aims to balance the load among multiple controllers by assigning switches to controllers. By considering both reactive and proactive assignment in a multi-controller setting, two costs are studied. The bi- objective function is composed of the cost of load balancing within controllers and the cost of traffic load migration. The problem of controller assignment is formulated as a Quadratic Programming, constrained by computing resources. Finally, to overcome the curse of dimensionality due to the increasing number of variables, a min-max model is presented as a mixed-integer linear programming problem minimizing the maximum load of controllers. Simulation results shed light on the trade-off between load balancing and migration cost, and the performance evaluation is demonstrating the efficiency of the proposed model compared to previously proposed algorithms in the literature.

View graph of relations

© 2018 King's College London | Strand | London WC2R 2LS | England | United Kingdom | Tel +44 (0)20 7836 5454