论文部分内容阅读
日益增长的数据中心冷却系统能耗会提升数据中心运行成本并带来巨大的碳排放,在数据中心轻载时进行负载聚合是降低系统能耗的有效途径.传统负载聚合只关注计算功耗的降低,而负载聚合会带来数据中心热点的产生和冷却功耗的增加.通过热力学仿真得到温度功耗模型;以一个实际云计算平台内的工作日志作为负载进行数据中心仿真,结果表明考虑节点热传递特性的聚合算法比起考虑迁移开销的聚合算法更能有较降低峰值温度和冷却功耗.
Increasing data center cooling system energy consumption increases data center operating costs and leads to significant carbon emissions, while aggregating data at light loads at the data center is an effective way to reduce system energy consumption.Traditional load aggregation focuses on computing power consumption And the load aggregation will lead to the increase of data center hotspot and the increase of cooling power.According to the thermodynamic simulation, the temperature power model is obtained, and the working log in a real cloud computing platform is used as the data center simulation. The results show that considering the node Aggregation algorithms for heat transfer characteristics reduce the peak temperature and cooling power consumption more than aggregate algorithms that consider migration costs.