Towards a robust and green internet backbone network

Networking infrastructure consumes a sizable fraction of the electricity supply. A network design model aimed at maximizing energy savings by aggregating traffic demand at a small set of resources, to put under-utilized resources to sleep, is offset by legacy models aimed at maximizing the network throughput by spreading the load across network resources. Traffic fluctuations and sudden spikes further complicate the problem. In this paper, we show that the problem is NP-hard, and propose a heuristic which targets optimal power usage for common traffic demand, while accommodating traffic fluctuations. The efficacy of our design is tested on two published backbone networks: NLR and NSFNET. Simulation results reveal that the proposed heuristic closely matches the results of the optimal algorithm in both energy savings and network utilization, while enjoying polynomial time complexity.