随着大数据时代的到来，学术界和企业界普遍采用MapRedce集群支持大数据管理和分析，且因为可再生能源环境和经济成本低廉，人们更倾向于使用可再生能源供电的MapReduce集群。然而可再生能源通常具有间歇性、不稳定性和动态变化性，其生产的绿色电能的消纳一直是一个难题。在MapReduce集群中，该问题同样存在：集群耗电或小于绿色电能，产生大量浪费，或大于绿色电能，需要传统电力补足，究其原因是产电和耗电模型的不匹配造成的。本项目研究可再生能源供电的MapReduce集群能耗调优技术，给定可再生能源特征、硬件环境和任务属性等上下文，通过功耗限制、资源分配和任务管理等方法，在上下文不变的前提下调整MapReduce集群能耗，使集群满足在一定时间内“减少绿色电能的浪费，降低传统电能的消耗”的双重目标。本项目将研究成果集成至开源Hadoop MapReduce集群，并使用真实的风场产电数据和电商网站

With the explosive growth of data, clusters with support MapReduce platform have been widely adopted in both industry and academia to deal with big data managements and analysis. Considering its low environment and economic cost, it is better to adopt renewable energy to generate green power for MapReduce clusters. However, renewable energy is intermittent, instability and dynamic, which bring great challenges for utilizing green power. In a MapReduce cluster, green power is wasted when it is generated more than cluster’s consumption, and traditional power is required as supplement when green one is generated less than cluster’s consumption. The unmatched power generation model and power consumption model is the key reason. In theis project, we study energy consumption tuning techniques of MapReduce cluster powered by renewable energy. Given characteristics of renewable energy, a hardware environment, task attributes and other contexts, the bi-objective optimization of “improving the renewable energy utilization and reducing the traditional energy usage of MapReduce cluster” can be achieved in a time duration through energy consumption tuning, which is based on power limitation, resources allocation and tasks management, on the premise of invariant context. Eventually we will integrate our approaches to Hadoop MapReduce, and further implement a prototype system. Meanwhile, we will evaluate our approaches in the context abstracted from the power generation logs of wind power plant and database logs of e-commercial website.