Incremental Deployment of Hybrid IP/SDN Network with Optimized Traffic Engineering

Abstract

The recent introduction of Software Defined Networks (SDN) into the traditional networking paradigm to create hybrid SDN networks brings with it several economical, technical and organizational challenges which must be addressed. In deploying hybrid SDN networks, the maintenance of numerous factors is taken into consideration such as throughput, network traffic, load balancing and fast failure recovery. One strategy that has been suggested is the incremental deployment of SDN controllers alongside legacy networking systems to reap the benefits of both paradigms while minimizing disruptions to networking services and maintaining network performance from the perspective of traffic engineering. In this paper, we seek to explore an optimal incremental deployment sequence of legacy networking devices to programmable SDN switches based on traffic engineering measures, namely minimizing maximum link utilization, thus determining the most suitable devices to migrate. A combination of two metaheuristics algorithms (Particle Swarm Optimization and Ant Colony Optimization) are implemented to identify this optimal sequence in terms of the locations of routers to be migrated along with the optimal weight setting and flow split ratios at each stage of migration. The deployment sequence is simulated and compared with static migration algorithms for evaluation.