CSR: Small: Scheduling Energy Consumption in Green Datacenters

点击翻译按钮获取中文摘要

The massive energy consumption of today?s datacenters translates into high monetary and environmental costs; the latter because most of the electricity produced in the US comes from burning coal, a greenhouse-gas-intensive approach for producing energy. We refer to such energy as ?brown?, as opposed to the ?green? energy produced by clean energy sources.An increasingly popular approach for reducing both costs is for datacenters to generate their own green energy or draw power directly from a nearby green energy plant. In light of this trend, the goal of this project is to study how best to exploit solar and wind energy for lowering energy costs and brown energy consumption in datacenters. The major challenge in using these types of green energy is that they are not always available.In this context, our research focuses on: (1) Characterizing and modeling datacenter workloads and green energy production; (2) Designing load scheduling and energy usage policies; (3) Designing energy management policies that account for green energy; and (4) Designing systems that leverage our models and policies.Our project will impact society in many ways, including: (1) reducing the brown energy consumption and the carbon footprint of datacenters; (2) promoting the generation and consumption of green energy; and (3) creating the machinery needed to exploit green energy in datacenters for highest benefit. Furthermore, we believe that undertaking work with such clearly defined societal impact will help us attract and train a diverse and committed set of undergraduate and graduate students.