软件定义容器式云存储的自治管理关键技术研究

Software defined cloud storage system based on containers embodies the trend of the combination of software definition and container technology, providing efficient support for the operation and management of cloud applications. With autonomous management requirements for large-scale and complex cloud storage systems, this project is to explore the technical issues of software defined storage integrated with container virtualization technology. Based on the framework of Software defined container-style cloud storage system，heterogeneous storage resources and computing service capabilities are clustered, managed and scheduled by encapsulating into containers. The studies on the model, mechanism, technique and algorithm of autonomous management of container storage system from the software defined control plane.In order to meet the requirements of automatic container construction and Cooperation, three kinds of technology will be researched, including on-demand adaptation technology for heterogeneous storage system, adaptive context perception, and automatic container deployment. Both self-evolution and self-recovery of storage container are to be studied based on Perceiving context or Fault warning. The research results of the project will be verified by software development. The project work is expected to provide reasonable theoretical and technical foundations for the application and development of software defined computing.

软件定义的容器式云存储系统体现了软件定义计算和容器技术相融发展的特征，可为云计算系统的运行和管理提供高效的基础设施支持。面向大规模、复杂化的云存储系统的自治管理需求，把握软件定义计算和虚拟计算技术发展的潮流，探索软件定义存储与容器虚拟化融合应用的技术问题，构建软件定义的容器式云存储框架，通过对异构存储资源和计算服务能力进行容器化聚合、管理和调度，从软件定义的控制平面聚焦研究容器式存储系统自治管理的相关模型、机制、技术和算法。面向存储资源容器自动构造和协同的需求，研究面向异构存储的容器按需适配技术、应用情境自适应感知技术、容器自动部署技术；支持感知情境变化或故障预警的存储容器服务能力自演化和自恢复机制，并完成研究成果的软件系统验证，为软件定义计算系统自治管理的应用和发展提供相关的理论和技术基础。

面向软件定义计算和云存储系统融合的需求，本项目研究了软件定义的容器云存储系统的关键技术和实现机制。通过对异构存储资源和计算服务能力进行容器化封装和管理，聚焦研究容器云存储系统自治管理的相关模型、机制、技术和算法。研究存储资源容器自动构造和协同机制，提出面向异构存储资源的容器按需适配和自适应调度、动态运行时情境的自适应感知和数据处理优化、基于应用需求与存储资源映射的容器自动部署等解决方案。为支持容器服务能力动态演化，提出基于图模型的容器协同编排策略、基于纠删码的软件定义存储系统数据协同更新与修复机制、QoS感知的容器弹性调度机制。面向容器云存储系统的高可用性需求，提出容器故障主动预测和诊断方法、基于情境建模的灰色故障检测和预测、基于故障关联模型的容器级联故障检测策略、服务故障根因定位和故障恢复机制等。研发相关的软件系统进行研究成果的验证和应用。本项目的研究成果为云计算和软件定义技术的智能化发展和应用提供有益的理论和技术基础。