A data center energy efficiency optimization method based on optimal temperature control of designated active servers

• Published in: Energy and buildings Vol. 345; p. 116126
• Main Authors: Xu, Yizhe, Tian, Maoyu, Yan, Chengchu, Li, Depeng, Deng, Tao, Guo, Puxi, Hu, Kai
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 15.10.2025
• Subjects: Data center; IT and cooling systems; Joint scheduling; Optimal energy efficiency; Joint scheduling; Data center; Optimal energy efficiency; IT and cooling systems
• ISSN: 0378-7788
• DOI: 10.1016/j.enbuild.2025.116126

[Display omitted] •A feasible path achieves optimal energy efficiency for IT and cooling systems.•Due tothermoelectric coupling, the energy optimization problem has been converted.•Achieves coordination of energy-saving scheduling between IT and cooling systems.•Our method cuts energy by 13.6 % vs. conventional, via joint IT-cooling scheduling.•VM deployment frequency during load growth affects optimization’s energy savings. The proportion of energy consumed by IT and cooling systems in data centers surpasses 90 %; however, due to the substantial disparities in scheduling frequency, speed, and precision between these two systems, local hotspots and excessive cooling frequently arise during energy-saving scheduling, significantly affecting overall energy efficiency. The underlying cause is the lack of effective coordination between IT and cooling systems in energy-saving scheduling. Consequently, this study proposes a method for optimizing data center energy efficiency, focused on optimal temperature control of designated active servers. The core concept involves reframing the intricate control optimization challenge of “overall energy efficiency optimal” into the more manageable task of “optimal temperature control of active servers subject to given quantity constraints.” This approach achieves effective coordination between the energy-saving scheduling of IT and cooling systems via a division of labor, ultimately improving the data center’s overall energy efficiency. In this study, the proposed optimization method was implemented in a liquid-cooled data center in the Nanjing region. Actual load testing demonstrated that, compared with conventional energy-saving strategies, the optimization method proposed in this study can further decrease the total energy consumption of data centers by 12.09 % (homogeneous VMs) and 29.1 % (heterogeneous VMs) through the joint scheduling of IT and cooling systems, even when maintaining the same number of active servers. Furthermore, as the migration time interval diminishes, the energy-saving effect can be further enhanced.