Reducing Energy Costs in Internet-Scale Distributed Systems Using Load Shifting

Abstract—Energy costs have become a significant fraction of the operational costs of running large Internet-scale distributed systems (IDS). In this work, we propose a demand-response technique where the system temporarily reduces its energy usage in response to pricing signals from a smart grid. Our proposed demand-response technique involves deferring the load from elastic requests to later time periods in order to reduce the server demand and the current energy usage, and hence, energy costs. We propose an optimal offline algorithm for demand response and evaluate it on production workloads from a commercial content delivery network using realistic electricity pricing models. Our optimal offline algorithm can achieve 12 % energy cost savings for time-of-use electricity pricing, even when only 40 % of the load is elastic and the service delay is at most 6 hours. The savings increase to 32 % under peak-based demand pricing and to 23% under a combination of time-of-use and demand pricing. Further, we show that almost all the energy cost savings can be achieved with no increase in the bandwidth cost of the IDS. I