算力网络资源协同调度探索与应用

摘要：算力网络融合了多区域、多层级的算力资源。针对面向用户差异化需求下全域大范围多层次算力资源的弹性灵活调度问题，设计了一个算力网络资源协同调度平台。针对差异化业务需求，该平台可以形成多公有云与私有云之间的算力资源调度策略以及算网资源协同调度策略，实现资源的自动最优分配。同时，该平台能够自动发现、纳管算力资源，并基于新的资源池情况实现分配策略的自动调整，实现用户无感业务扩缩容，以此弹性调度算力资源。

关键词：算力网络；协同调度；微服务；动态扩缩容

Abstract: The computing power network integrates computing power resources from multiple regions and levels. In response to the flexible scheduling problem of computing power resources across a wide range and at multiple levels under user differentiated needs, a computing power network resource collaborative scheduling platform is designed. The platform forms a computing power resource scheduling strategy between multiple public and private clouds and a computing network resource collaborative scheduling strategy based on differentiated business needs, to achieve automated and optimal allocation of corresponding resources. At the same time, the platform can automatically discover and manage computing resources, and automatically adjust allocation strategies based on the new resource pool situation, achieving user insensitive business expansion and contraction, thereby flexibly scheduling computing resources.

Keywords: computational power network; collaborative scheduling; micro services; dynamic scaling