运营商行业低碳路径探索:数据中心的智能化调度策略研究  

Exploring Low Carbon Paths in the Carrier Industry:Research on Intelligent Scheduling Strategies for Data Centres

作　　者：李子然 熊国华 史磊 胡燕妮 Li Ziran;Xiong Guohua;Shi Lei;Hu Yanni(China Railway Fifth Survey and Design Institute Group Co.,Ltd,Beijing 102627,China;Hunan Railway Construction and Investment Co.,Ltd,Changsha 410013,China;DTCT Data&Technology Co.,Ltd,Ningbo 315000,China)

机构地区：[1]中铁第五勘察设计院集团有限公司,北京102627 [2]湖南铁路建设投资有限公司,长沙410013 [3]浙江德塔森特数据技术有限公司,宁波315000

出　　处：《智能建筑电气技术》2025年第1期34-37,41,共5页Electrical Technology of Intelligent Buildings

摘　　要：针对运营商数据中心高能耗、热管理效率低等问题,本文提出一种基于ETL过程的算力-热力协同智能调度策略。通过构建多数据库环境下的数据流动模型,揭示算力需求与冷却系统热负荷的动态耦合关系,并建立融合实时热力响应的能量管理模型。研究表明,该方法通过动态优化算力分配与冷却调度,较传统策略可降低17.3%的PUE值,在保持服务质量的前提下实现年碳排放量减少21.5%。实验验证表明,基于设备运行参数自适应的ETL协同机制能有效缓解局部热点现象,服务器集群温度波动幅度降低至±1.2℃。该模型为运营商数据中心低碳转型提供了可行的智能化解决方案。Aiming at the problems of high energy consumption and low thermal management efficiency in operators'data centres,this paper proposes an arithmetic-thermal cooperative intelligent scheduling strategy based on the ETL process.By constructing a data flow model in a multi-database environment,it reveals the dynamic coupling relationship between arithmetic demand and cooling system thermal load,and establishes an energy management model that incorporates real-time thermal response.It is shown that the method can reduce the PUE value by 17.3%compared with the traditional strategy by dynamically optimising the arithmetic allocation and cooling scheduling,and achieve a 21.5%reduction in annual carbon emissions while maintaining the quality of service.Experimental validation shows that the ETL cooperative mechanism based on adaptive equipment operating parameters can effectively mitigate the local hotspot phenomenon,and the temperature fluctuation of the server cluster is reduced to±1.2℃.The model provides a feasible intelligent solution for the low-carbon transformation of operator data centres.

关 键 词：数据中心 低碳路径 智能化调度 算力-热力协同 多数据负载 

分 类 号：TP308[自动化与计算机技术—计算机系统结构]