SDN based Multi Path Routing for Efficient Cost Minimization - Advanced Networking Algorithm

Abstract

Software Defined Networking (SDN) is introduced to improve network programmability, simplify network management and enhance network operation. It is a paradigm shift in networking that suggests to decouple the control plane from the data plane and place the control of the system in a logically centralized node called „SDN controller‟. The existing system, examine the problem where the optimality gap of iterative routing algorithms decreases exponentially fast and we want to minimize the average routing cost subject to a constraint for the average reconfiguration frequency. Furthermore, we present two control policies working on top of the online routing optimization engine to decide whether to apply or not the current yet not optimal global network configuration. Numerical results on the GEANT network and fat-tree network topologies show that our control schemes can effectively track the evolution of the system using a bounded number of reconfigurations, thus pursuing the double objective of optimizing the performance and his system stability. The recent technology on routing has given many solutions that it is carried out to many SDN networks, solving online versions of the problem with computing consistent flow migration. This thesis presents an autonomous network reconfiguration system (ARS) that enables a SDN to autonomously recover from local link failures to preserve network performance. By using ARS generates necessary changes in routing and algorithm with channel assignments in order to recover from failures.