Carbon-aware routing in software defined inter data center network

Abstract

Data center operators are trying to reduce environmental cost (carbon emission), by establishing data centers at locations where renewable energy (wind or solar energy) is available. To maximize the use of renewable energy the connection-request is routed to these “Green” data centers rather than “Brown” data centers (powered by hydrocarbon energy), subject to content availability, latency to the end-host and load at that data center. However availability of renewable energy varies throughout the day and is subjected to weather conditions. Such carbon-aware routing require the global knowledge of renewable energy availability at green data centers in the network. Centralized control can help in optimizing network-wide energy use. This paper investigates the effect of carbon-aware routing in OpenFlow based software defined inter-data center optical-network (wherein some data centers powered by green energy), on energy cost, energy consumption and carbon emission in the network. Green Data centers have higher energy cost compared to brown data centers as production of green energy is costly compared to brown energy. We propose priority-based carbon-aware routing scheme and compare its performance in terms of energy cost and carbon emission with two scheme namely Nearest and Green. Numerical results obtained from simulation show that priority-based carbon-aware routing is optimum (with assumed emission factor and energy cost), in terms of trade-off between carbon emission and energy cost, when compared with other two routing schemes.