随着5G和工业物理网等新兴技术的发展，数据处理和计算需求不断增长，跨域数据中心级联已成为分布式系统的研究热点之一。现有多数据中心结构能实现资源整合和数据跨域存储，但网络跨域互联时带宽的动态变化导致了资源级联稳定性的下降及任务跨域调度时间的急剧增加，而且非原生网络跨域拓扑也易造成系统性能的下降。.鉴此，拟多数据中心动态网络下的资源协同为目标，基于OpenStack平台架构，网络跨域虚拟化、网络QoS等技术，设计面向动态网络的多数据中心资源级联体系结构。通过Neutron组件和网络资源虚拟化技术的综合运用，解决多数据中心L2/L3层虚拟网络跨域实现问题。最后，提出多数据中心高效实时资源分配算法，实现面向大规模计算任务的资源跨域时空调度。研究成果将推进多数据中心新型原生体系结构设计，还将有助于解决多数据中心数据安全及容错问题，具有较重要的理论与实际意义。

With the development of emerging technologies such as 5G and industrial internet of things, the demands for computing power and data processing are ever-growing. Cross-domain datacenter cascading has become the hot topic of the distributed system. Existing multi-datacenter architectures can basically achieve resource integration and data cross-domain storage, but the dynamics of bandwidth since networking cross-domain interconnecting leads to a surge in both the difficulty of resource cascading and scheduling time of cross-domain tasks. Non-native networking cross-domain topologies are also prone to degradation of system performance..For this reason, this project is targeted at the resource coordination in the multi-datacenter dynamic networking environment, and base on the OpenStack platform architecture, networking cross-domain virtualization and networking QoS, design a multi-datacenter resource cascading architecture oriented to the dynamic networking. With the combined use of Neutron component and networking resource virtualization technologies, solve the multi-datacenter virtual networking cross-domain realization problems on the L2/L3 level. Finally, we design a multi-datacenter high-efficient real-time resource allocation algorithm and implement a resource cross-domain space-time scheduling strategy oriented to the large-scale computing tasks. The research outputs will advance the novel multi-datacenter native architecture exploration, and also can be used to solve the application problems in data security and faulty tolerant in the multi-datacenter environment. It has important theoretical and practical value.