基于云架构的存储信息系统研究

发布时间：2018-05-16 16:45

本文选题：云计算 + 问题求解模型　；参考：《太原理工大学》2012年博士论文

【摘要】：云计算是一种支持共享的网络交付信息服务的模式,云服务的使用者看到的只有服务本身,而不用关心相关基础设施的具体实现。云计算的两个本质特征是资源虚拟化和处理并行化。资源虚拟化是对计算资源和存储资源的逻辑表示,它不受物理限制的约束。现有云存储架构技术的主要目标是解决如何在互联网环境下对云计算、云存储进行合理抽象,建立反映新一代存储系统主要特征的云存储系统架构模型。这些研究的共同点是把存储介质视为云存储的最小单元,云存储未穿透该最小单元,因此未形成存储介质并行处理的云架构。本文主要针对存储介质的结构开展研究,提出一种新型存储结构：基于云计算架构的存储信息系统(CBSIS)。CBSIS采用并行处理的云架构组织文件管理系统与存储介质,即在建立CBSIS元模型-问题求解模型(PRM)的基础上,将文件管理系统与存储介质映射为原子构件(AC)与PRM的引擎(E),用功能集成电路IP核实现AC的原子系统构件(AS),并部署成PRM并行树的叶节点,基于消息访问PRM树。最终形成云存储架构的片上系统(SSoC)。论文从四方面重点阐述三个创新性研究成果：研究基于云计算架构的存储信息系统模型；基于该模型分析设计和验证新一代CBSIS云存储系统；最终实现基于CBSIS云存储片上系统(SSoC)实物原型系统。 (1)通过可计算架构的问题求解模型(PRM),课题建立了基于PRM元模型的云存储系统架构模型CBSIS,形成本文的第一个研究成果。本研究提出了针对具体问题建立问题求解模型的方法,包括建立概念模型、逻辑模型和物理模型的三层SOA架构。物理模型映射为SOA的构件,使构件池的原子构件集合的有限集可以通过逻辑模型的映射,表现概念模型的语义无穷集。构件间是松耦合的,以消息的形式通过接口访问,构成云计算的资源虚拟化基础。 (2)CBSIS体系结构和总体框架的设计方案具有如下特点：(a)将体系结构引入CBSIS系统设计；(b)将引擎E与体系结构连接件CON相关联,使PRM的部署与CON的部署相一致。通过提出和定义CBSIS应用层语言LL7,并用连接件CON模型和构件AC模型实现LL7,形成本研究的第二个创新点。 (3)以生成CBSIS可执行原型系统为背景,讨论了可执行业务和可执行架构原型系统,并提出了增量式原型系统的概念,可用于系统分析、设计和实现的各个阶段。(4)本研究提出的云存储片上系统SSoC解决方案,其特点是针对云存储系统的现状及未来发展趋势,用于对云存储基础设施、管理平台直至整个数据中心的分阶段改造与部署实施。论文以云存储架构的基础设施改造为例,提出并实现本研究的第三个成果：基于SoW和bSATA控制器的SSoC。其SSoC是以集成文件管理系统模型和形成该模型要素的分层嵌套PRM树结构为特征的。本研究提出的CBSIS架构及其设计、验证和原型系统生成与实现方法是针对云存储应用的,但使用的方法和工具具有普遍性,不局限于具体应用领域。因此本研究提出的方法与理念适用于一般应用系统的建模,具有快速开发与部署、易于测试和易于维护等特征。
[Abstract]:Cloud computing is a pattern that supports the shared network delivery of information services. The users of the cloud service see only the service itself, without care about the specific implementation of the related infrastructure. The two essential features of the cloud computing are the virtualization of resources and the parallelization of processing. It is not bound by physical restrictions.
The main goal of the existing cloud storage architecture technology is to solve cloud computing and cloud storage in the Internet environment, and to establish a cloud storage architecture model which reflects the main features of the new generation storage system. Therefore, there is no cloud architecture for parallel processing of storage medium.
This paper focuses on the structure of storage medium, and proposes a new storage structure: the storage information system based on cloud computing architecture (CBSIS).CBSIS adopts the parallel processing Cloud Architecture to organize file management system and storage medium, which is based on the establishment of the CBSIS meta model problem solving model (PRM) and the file management system and the file management system. The storage medium is mapped to the atomic component (AC) and the PRM engine (E), and the AC's atomic system component (AS) is implemented with the IP kernel of the functional integrated circuit, and is deployed into the node of the PRM parallel tree, and the PRM tree is accessed based on the message. Finally, the system on chip (SSoC) of the cloud storage architecture is formed.
This paper focuses on three innovative research results from four aspects: Research on the model of storage information system based on cloud computing architecture, and design and verify a new generation of CBSIS cloud storage system based on this model; finally, it is based on the prototype system of the system on the CBSIS cloud storage chip (SSoC).
(1) through the problem-solving model (PRM) of computable architecture (PRM), a cloud storage system architecture model based on PRM metamodel is established to form the first research result of this paper. This study proposes a method for establishing a problem solving model for specific problems, including the establishment of a conceptual model, a logical model and a physical model of the three layer SOA architecture. The physical model is mapped to the component of SOA. The finite set of the set of atomic components in the component pool can be mapped by the logical model, and the semantic infinite set of the conceptual model is expressed. The component is loosely coupled, and is accessed through the interface in the form of message, which forms the basis of the resource virtualization of the cloud computing.
(2) the design scheme of the CBSIS architecture and the overall framework has the following features: (a) introducing the architecture into the CBSIS system design; (b) associating engine E with the architecture connector CON to make the deployment of PRM consistent with the deployment of CON.
By proposing and defining the CBSIS application level language LL7, and implementing the LL7 with the connector CON model and the component AC model, second innovative points of this research are formed.
(3) with the background of generating CBSIS executable prototype system, the executable service and executable architecture prototype system are discussed, and the concept of incremental prototype system is proposed, which can be used in the various stages of system analysis, design and implementation. (4) the cloud storage system on the cloud storage system is proposed in this study, which is characterized by the present cloud storage system. And future development trend for cloud storage infrastructure, management platform until the entire data center staged transformation and deployment implementation.
This paper, taking the infrastructure transformation of cloud storage architecture as an example, proposes and implements third results of this study: the SSoC of SSoC. based on SoW and bSATA controller is characterized by the integrated file management system model and the hierarchical nested PRM tree structure forming the elements of the model.
CBSIS architecture and its design, design, verification and prototype system generation and implementation are applied to cloud storage, but the methods and tools used are universal, not limited to specific applications. Therefore, the methods and concepts proposed in this study are applicable to the modeling of general application systems, with rapid development and deployment and easy to measure. Test and easy to maintain and so on.

【学位授予单位】：太原理工大学
【学位级别】：博士
【学位授予年份】：2012
【分类号】：TP333

【参考文献】