An online power-aware routing in SDN with congestion-avoidance traffic reallocation

Abstract

Software-Defined Networks (SDN) can be seen as a promising alternative to achieve the long-awaited power efficiency in current communications systems. In these programmable networks a power-aware mechanism could be easily implemented leveraging the capabilities provided by control and data plane separation. For such purpose, this paper proposes a novel solution minimizing the number of active elements required in an SDN with multiple controllers and in-band control traffic. In order to provide a complete and fine-grained strategy, this proposal comprises two crucial modules: GrIS, a green initial setup and DyPAR, a dynamic power-aware routing. Besides being compatible with SDN environments without a dedicated control network, the proposed strategy is able to handle demanding traffic arrival without degrading the performance of higher priority traffic. Simulation results show that our heuristic approach allows to obtain close-to-optimal power savings with differences under 8%. Moreover, comparison with existing related methods using real topologies validates the improvements achieved by our solution in terms of power efficiency and performance degradation avoidance. For instance, after routing all the incoming traffic, a reduction of power consumption of up to 26.5% and an increase of allocated demands of up to 26.7% can be reached by our solution.