多租户多数据中心的容器云平台虚拟资源调度研究

There is an uncertain complex nonlinear dynamic relationship between computing resources and network resources, the two main virtual resources in the cloud platform for the large-scale multi-tenant and multi-data center, which is based on the docker as being the new virtual unit. Thus, these two resources are adaptively allocated in a collaborative mode. Using high-performance machine learning algorithm and the multi-agent technology with the docker cluster as the unit, we probe into the co-adaptive tasks scheduling theory, the key technologies and the methods of the virtual computing and network resources in the multi-tenant and multi-data center under the cloud environment, so as to balance the maximum interests of both supply and demand of cloud services on the premise of meeting the User Service Level Agreements. In this study, we focus on: firstly, the adaptive tasks scheduling theory in terms of the computing resources and the networks resources respectively in the multi-data centers with the virtual docker technology, in order to satisfy the need of the multi-tenant services quality, including the selection of data centers, docker clusters and the server, and the adaptive scheduling mechanism of computing resources and networks resources in the docker cloud platform; secondly, the co-adaptive scheduling algorithm for virtual network resources in the docker cloud platform for the multi-tenant and multi-data centers, including the high-performance co-adaptive scheduling mechanism and the knowledge transfer mechanism; thirdly, the setup of a prototype system to verify the proposed theory, technologies and methods.

以容器为新型虚拟化单元的大规模多租户多数据中心云平台中的主要两类虚拟资源——计算资源和网络资源之间存在综错复杂的非线性动态不确定关系，客观上要求两者以协同方式进行自适应配置。本项目利用高性能机器学习方法和多agent技术，以容器簇为粒度，研究多租户多数据中心的云服务环境下虚拟计算资源和网络资源协同自适应调度的理论问题、关键技术和方法，以确保在满足用户服务等级协议前提下，最大程度实现云服务供需双方的利益均衡。主要研究内容包括：分别从虚拟化容器技术下多数据中心计算资源和网络资源两个层面研究满足多租户服务质量需求的自适应调度机制，包括数据中心、容器簇和服务器的选择，以及容器云平台计算资源和网络资源的自适应调度机制；研究多租户多数据中心环境下容器云平台计算资源与网络资源协同自适应调度机制，包括高性能协同自适应调度算法、知识迁移机制等研究；构建原型验证系统，验证本项目所提的理论、技术和方法。

以容器为新型虚拟化单元的大规模多租户多数据中心云平台中的主要两类虚拟资源——计算资源和网络资源之间存在综错复杂的非线性动态不确定关系，客观上要求两者以协同方式进行自适应配置。本项目利用高性能机器学习方法和多agent技术，以容器簇为粒度，研究多租户多数据中心的云服务环境下虚拟计算资源和网络资源协同自适应调度的理论问题、关键技术和方法，以确保在满足用户服务等级协议前提下，最大程度实现云服务供需双方的利益均衡。主要研究内容包括：分别从虚拟化容器技术下多数据中心计算资源和网络资源两个层面研究满足多租户服务质量需求的自适应调度机制，包括数据中心、容器簇和服务器的选择，以及容器云平台计算资源和网络资源的自适应调度机制；研究多租户多数据中心环境下容器云平台计算资源与网络资源协同自适应调度机制，包括高性能协同自适应调度算法、知识迁移机制等研究；构建原型验证系统，验证本项目所提的理论、技术和方法。