中文摘要：

针对大规模计算的能效问题,提出改进的能效度量模型,通过＂能源＂和＂效率＂2种度量来综合评价系统能效.在＂能源＂方面,考虑计算机、网络和附属设备的能耗;在＂效率＂方面,考虑CPU、内存、磁盘以及网络的情况.提出的能效度量模型描述了改进后的能效度量的定义和数学表达,通过实验验证了该模型的合理性.基于该度量模型,评估并分析了MapReduce环境中CPU密集型、I/O密集型和交互型计算的能效,总结了MapReduce环境中的能效规律.

英文摘要：

An energy efficiency measurement model was proposed to address the large-scale computing problem through two metrics： ＂energy＂ and ＂efficiency＂. As for ＂energy＂, the energy consumption of computer, network and affiliated equipments was considered; as for ＂efficiency＂, that of CPU, memory, disk and network was considered. The proposed energy efficiency measurement model describes the defini- tion and mathematical expression of the improved energy efficiency measurement, and is proved reasonable through experiments. The energy efficiency of CPU intensive, I/O intensive and interactive computing was evaluated and analyzed based on the measurement model, and the energy efficiency laws in MapReduce environment were summarized.