A fuzzy logic-based controller for cost and energy efficient load balancing in geo-distributed data centers

The ever-increasing demand for cloud services results in large electricity costs to cloud providers and causes significant impact on the environment. This has pushed cloud providers to power their data centers with renewable energy sources more than ever. Among the different ways of adopting renewable energy sources, on-site power generation using wind and solar energy has gained considerable attention by large companies and proved its potential to reduce data centers' carbon footprint and energy costs. Efficient utilization of renewable energy sources is challenging due to their intermittency and unpredictability. Cloud providers with multiple Geo-distributed data centers in a region can exploit the temporal variations in on-site power and grid power price by routing the load to a suitable data center in order to reduce cost and increase renewable energy utilization. To achieve this goal, we propose a fuzzy logic-based load balancing algorithm that acts with no knowledge of future. We conduct extensive experiments using a case study based on real world traces obtained from National Renewable Energy Laboratory (NREL) and Energy Information Administration (EIA) in the US, and Google cluster-usage. Compared to other benchmark algorithms, our method is able to significantly reduce the cost without a priori knowledge of the future electricity price, renewable energy availability, and workloads.