摘要
随着科技发展和数字经济的迅速崛起,算力网络逐渐成为优化和利用分布式计算资源与网络资源的重要手段。本文旨在探讨分布式算力网络资源建模和调度优化方法。在算力网络中,资源分配算法的核心在于如何根据业务需求优化计算资源及网络资源的智能调度。首先,本文深入研究了算力网络的定义、基本架构及其关键技术。其次,对比分析了目前算力网络中资源分配算法的特性,特别是基于边缘计算和算力网络建模的资源分配方法。这些算法旨在通过优化算力资源的分配,提高系统的整体性能,满足不同业务需求下的计算任务。最后,论述了算力网络管理技术的发展趋势。
With the rapid development of technology and the rise of the digital economy,computing power networks have gradually become an important means of optimizing and utilizing distributed computing resources and network resources.This article aims to explore the modeling and scheduling optimization methods of distributed computing network resources.In computing power networks,the core of resource allocation algorithms lies in how to optimize the intelligent scheduling of computing and network resources based on business needs.First,this article delves into the definition,basic architecture,and key technologies of computing power networks.Second,the characteristics of the current resource allocation algorithms in computational power network are compared and analyzed,especially the resource allocation methods based on edge computing and computational power network modeling.These algorithms aim to optimize the allocation of computing power resources,improve the overall performance of the system,and fullfil the computing tasks under different business requirements.Finally,the development trend of computing power network management technology is discussed.
作者
李俊菲
曾庆涛
陆利坤
李文璟
LI Junfei;ZENG Qingtao;LU Likun;LI Wenjing(School of Information Engineering,Beijing Institute of Graphic Communication,Beijing 102627;National Key Laboratory of Network and Switching Technology,Beijing University of Posts and Telecommunications,Beijing 100876)
出处
《高技术通讯》
北大核心
2025年第12期1337-1350,共14页
Chinese High Technology Letters
基金
网络与交换技术全国重点实验室(北京邮电大学)开放课题(SKLNST-2023-1-12)
北京市高等教育学会2023年立项面上课题(MS2023168)
北京市教育委员会出版学新兴交叉学科平台建设(04190123001/003)资助项目。
关键词
算力网络
边缘计算
算力资源分配
算力网络的应用
computing power network
edge computing
allocation of computing power resources
the application of computing power network
