The demand for computing power is growing rapidly in the digital economy,and digital transformation of enterprises depends on computing power resources and related applications.Based on the analysis of the four major categories of computing power applications in the entire process of computing power manufacturing,a system architecture for computing power networks of enterprise campus is proposed.Based on this,four subsystems covering research and development cloud,office cloud,production cloud,and terminal cloud are constructed,and a computing power network system for enterprise campus is established around the three major technological innovations.This paper also proposes for the first time that the core competitiveness of computing power manufacturing is equivalent to the efficiency and cost of using computing power in these applications.Only by implementing dual customization in resources and operations can we achieve the ultimate efficiency of resource utilization and the ultimate experience of operational services,thereby achieving the goal of cost reduction and efficiency improvement in computing power manufacturing.The system and key technologies were pilot verified in the cooperation projects between operators and enterprises.The results show that significant achievements have been made in improving the application experience and computing resource efficiency,becoming a key factor in the competitiveness of computing servers.
关键词
算力制造/算力调度/算力并网/云边端协同/算力路由/企业广域网/业务优先级调度/业务服务质量
Key words
Computing Power Manufacturing/Computing Power Scheduling/Computing Power Grid Connection/End-Edge-Cloud Collaboration/Computing Power Routing/Enterprise WAN/Service Priority Scheduling/QoS
维普
万方数据