Minimizing Electricity Cost and Emissions in Optical Data Center Networks

Abstract

Information and communication technology plays an important role in worldwide electricity consumption. Existing routing approaches, oblivious to location-dependent energy prices and emissions, incur considerable costs and environmental impacts to network operators. We propose a new routing approach in order to minimize the electricity costs and emissions of core networks under multiple electricity market environments. The proposed approach is based on the advantages of using adaptive routing. We use the available electricity price and fuel mix from power utilities to find the lowest electricity cost path and the least emissions path, while the electricity price and emissions depend on the location diversity and time diversity. Depending on the electricity price and emissions, alternative geographical paths in wide-area networks may be preferred. We propose an analytical model for electricity costs and emissions that considers the blocking probability of network traffic on optical wavelength division multiplexing (WDM) networks with no wavelength conversion. We evaluate new approaches on two realistic topologies, and the results show up to a 26% improvement in electricity cost and 5% in emissions compared with two fixed routing approaches. Our balanced results imply that the proposed approach would be able to trade off between the electricity cost and emissions.