面向算力调度的省域有线传输链路调度优化研究
Research on Scheduling Optimization of Provincial Wired Transmission Links Oriented Towards Computing Power Scheduling
当前,该省域算力网络普遍存在链路负载失衡、传输时延偏高、资源利用率低、算网协同能力弱等问题影响了全域算力调度效率.因此,本文结合省域算力业务特点搭建算网协同调度架构,综合考虑带宽、时延等约束,采用多目标优化模型求解,并使用改进遗传算法求解,克服以往传统算法易陷入局部最优、应付突发任务能力弱等缺点.仿真结果表明,可有效平衡网络链路负载、缩减网络传输时延、提升网络资源利用率及调度稳定性,可为省域算力网络传输优化与算力协同调度提供可行的技术参考.
Currently, the provincial computing power network generally faces issues such as imbalanced link loads, high transmission delays, low resource utilization rates, and weak computing-network coordination capabilities, which affect the efficiency of global computing power scheduling. Therefore, this paper builds a computing-network coordinated scheduling architecture based on the characteristics of provincial computing power services, comprehensively considers constraints such as bandwidth and latency, employs a multi-objective optimization model for solution, and uses an improved genetic algorithm to overcome the shortcomings of traditional algorithms, such as being prone to local optima and weak ability to handle sudden tasks. Simulation results show that it can effectively balance network link loads, reduce network transmission delays, improve network resource utilization rates, and enhance scheduling stability, providing a feasible technical reference for the transmission optimization and computing power coordinated scheduling of provincial computing power networks.
| [1] |
唐卓, 蒋冰婷, 张嘉鹏, |
| [2] |
李逸博, 李小平, 王爽, |
| [3] |
刘运奇, 邓玲, 冯铭能, |
| [4] |
杨振东, 冯铭能, 陈善杰, |
/
| 〈 |
|
〉 |