面向大数据的绿色IT框架能效分类机制

摘要

与分散处理相比较，大数据中心集中处理信息通信任务，在能效上已有巨大的提高。但大数据中心包括数以千万记的服务器，其能源消耗量甚至可以超过一座小型城镇。巨大的能源消耗、成吨的温室气体排放，使大数据中心在能效与减排方面面临诸多挑战，建立绿色高效的大数据中心势在必行。本文给出一个面向大数据的绿色 IT 框架，重点研究了能效分类问题，提出了一个基于度量模型的能效分类机制。根据工作量和能源消耗情况对设备和服务进行分类，无缝地划分为不同的资源池，将电力使用效率、数据中心工作效率和二氧化碳排放等综合计算衡量，制定可以实现并遵循的能效标准，使绘制大数据中心的碳足迹成为可能，并提供服务能效评估方案。%Compared with the distributed processing, large data center focuses on information and commu-nication tasks and has already made huge improvement on energy efficiency. But large data center includes tens of millions of servers, and its energy consumption can be ever larger than one small town. Huge energy consumption and tons of greenhouse gas emission bring many challenges to large data center in terms of energy efficiency and emission reduction. Therefore, the establishment of green and efficient large data center is imperative. This paper provides a framework of green IT for large data, focusing on energy efficiency classification and proposes an en-ergy efficiency classification mechanism based on measurement model. According to the workload and energy consumption, this paper classifies the equipment and services seamlessly into different resource pool. Power effi-ciency, data center efficiency and carbon dioxide emissions are calculated and measured. This paper also formu-lates energy efficiency standards which could be realized and followed. The standards make the drawing of the big data center carbon trace possible and provide scheme for energy efficiency evaluation.