【作者】 周大海；

【导师】 夏秀峰；

【作者基本信息】 沈阳航空工业学院 ， 计算机应用技术， 2008， 硕士

【摘要】 许多组织机构建立了能够满足某个领域或某个部门需求的数据仓库系统。但随着企业规模的不断扩大及企业的合并与重组,这些在不同时期建立的独立数据仓库系统的弊病逐渐显现出来,因而出现了独立演进型分布式数据仓库技术。分布式数据仓库把分散的数据有机整合起来,使决策者能从大量数据中获得有价值的信息,以便更好地支持决策行为。本文围绕独立演进型数据仓库的体系结构、一致性维护策略、分布式ETL等关键技术进行了深入细致地研究,结合实际项目,提出了基于文件传输的数据一致性维护策略。本文重点研究了分布式数据仓库的数据抽取策略和网络环境下的数据传输策略,提出了一种通用的基于文件的传输数据构造形式和数据传输的实时规则、适时主动规则、适时被动规则及非定时规则。在全局数据文件字典和中心节点数据文件字典的支持下,提出了正常传输、定时传输和后台传输三种策略,较好解决了数据仓库在分布式环境下的数据集成问题。并将这些关键技术应用于特种设备监督检验与安全监察数据仓库系统的设计与开发中,获得了较好的效果。由于在分布式环境下数据的抽取、转换、装载是一个非常复杂的问题,因此与传统集中式环境下基于“ETL引擎”的方式有较大不同,本文仅对分布式ETL的概念模型进行了研究。更多还原

【Abstract】 Many data warehouse systems are established for a certain area by many organizations. As the expanding, merger and reorganization of the enterprises, the weaknesses of the independent data warehouse systems established at different times are emerged gradually, so the independent evolution of distributed data warehouse technology is appered. The scattered data is integrated by distributed data warehouse, so valuable information is obtained from large amounts of data by decision-makers in order to support the decision-making better. The architecture, consistency maintenance strategy, distributed ETL and other key technologies of the independent evolution of data warehouse are introduced by the paper, and a new data consistency maintenance strategy based on files are proposed in combination with the actual project.The extracting, transmission strategy of the data of the distributed data warehouse under network environment are researched by the paper, a universal constitution form of transmission data and the real-time rule, timing active rule, timing passive rule and non-timing rule of data transmission are proposed. By using the global data file dictionary and center node data file dictionary, three strategies of normal transmission, timing transmission and background transmission were advanced, so the issue of data integration in data warehouse can preferably solved in computer network environment. The key technologies are used on the designing and developing data warehouse system of supervision, inspection and security for special equipments, and good effects are obtained.The problems of the data extraction, transformation, loading of data under distributed environment are complicated, so they’re different from the mode of“ETL engine”based on traditional centralized environment. The conception mode of distributed ETL is investigated in the paper.更多还原