Research Progress on Computing Resource Scheduling in Computing Power Network
Songling Tang, Congfeng Jiang, Long Yan, Zhipeng Feng, Yonghe Guo, Jianing Niu, Chunpeng Wu, Xingping Wu, Xinyi Cai, Junming Liu · 2024
In the evolution of computing power networks, the effective scheduling of computing resources has become a critical concern, driven by the diversification of computing demands and the increasing heterogeneity of resources. This paper provides a review of scheduling strategies for computing resources within these networks, analyzing methods that are relevant for the current landscape, including CPUs and GPUs. It emphasizes application scenarios and benefits across three key areas: improving resource utilization, balancing fairness and efficiency in scheduling, and minimizing job completion time. By synthesizing and evaluating existing methods, this paper identifies potential future research directions in computing resource scheduling, including the effective scheduling of heterogeneous resources, joint scheduling of multiple resources, cross-domain scheduling, and optimization of energy consumption. Furthermore, the paper addresses the significance of tackling future challenges in computing resource scheduling, focusing on three aspects: heterogeneous perception and measurement, large-scale multi-resource collaborative scheduling, and cross-domain communication.