A Data Load Spatio-Temporal Scheduling Method Considering Thermal Inertia in Data Centres

With the rapid development of artificial intelligence and big data technologies, the continuous expansion of the scale of data centers has led to a significant increase in power load and energy consumption. Traditional scheduling methods are no longer sufficient to meet practical demands. This paper proposes a spatio-temporal scheduling approach for data loads, based on the energy consumption characteristics of IT equipment and cooling systems in data centers. The approach fully utilizes the potential flexibility of data loads through delayed processing and spatial transfer, and constructs a spatio-temporal scheduling model for data loads in data centers. Four strategies are explored in the model, namely, the basic strategy, the strategy considering delay, the strategy considering transfer, and the joint scheduling strategy of the two, and are verified through three typical data center scenarios located in different regions. The experimental results show that the joint spatio-temporal scheduling strategy exhibits significant advantages in mitigating transient load peaks, reducing overall energy consumption and improving system resource utilization, providing a practical solution for data center energy management.