基于XMLA标准的数据仓库统一访问接口的研究与实现

【作者】 刘志；

【导师】 游进国；

【作者基本信息】 昆明理工大学 ， 计算机应用技术， 2012， 硕士

【摘要】 随着数据库技术的发展和数据库系统的广泛使用,各种各样的数据库已经积累海量数据。联机分析处理(Online Analytical Processing以下简称OLAP)能从这些海量数据中寻找关联规律和发现潜在的有用信息作为企业决策依据,它主要在数据仓库基础上进行数据分析以求获得有用信息。从另一方面来看,支持高级决策OLAP技术可以深入分析和研究数据、发现趋势、看到异常情况、得到重要细节以便更好地理解商业行为的涨落变迁,已经得到广泛应用。但是,很多OLAP厂商都有内置数据访问标准而没有统一的数据访问标准,这给数据仓库统一访问甚至集成带来很大挑战。为了达到异构数据仓库之间能够通讯甚至集成的目标,很多生产OLAP工具的厂商努力开发能够访问所有OLAP服务器的统一OLAP API访问接口。可以从三个角度分析开发访问OLAP服务器统一OLAP API标准的必要性：(1)从数据仓库开发者角度,单一工具很少能完全满足用户不断变化的需求,又很难对各种产品进行集成。(2)从数据仓库用户角度,面对海量数据,无法轻易找到自己真正需要的信息,而且把这些信息完整正确地表示出来也是个挑战。(3)从数据仓库供应商角度,如果信息共享没有标准格式,元数据集成代价太大。本文首先简单介绍国内外研究现状以及数据仓库、OLAP技术相关理论,接着了解微软OLAP服务器元数据模型、Mondrian元数据模型和Oracle OLAP服务器元数据模型为实现各种数据仓库元数据标准的统一访问奠定理论基础。着重研究数据仓库三种访问标准以及MDX查询语言并对XMLA标准和MDX查询语言进行详细分析,因为Oracle OLAP服务器不支持XMLA标准和MDX语言。若实现符合XMLA标准的数据仓库统一访问的目标,必须首先实现Oracle OLAP元数据模型与XMLA数据模型的映射,然后根据MDX语言特点开发MDX解析器来实现对MDX语言的解析,根据以上理论知识设计出系统总体架构,该架构采用B/S三层结构,前端展示层使用开源JPivot产品,该产品是一个自定义JSP标签库,基本涵盖所有OLAP功能和操作,并且支持以XMLA标准访问数据源：控制层利用开发的符合XMLA标准的引擎(以下简称XMLA引擎)；数据层利用Oracle11g数据库。其次,设计并实现Oracle多维数据库为XMLA引擎提供数据来源。最后,实现系统并对系统进行测试,前台JPivot通过XMLA引擎对Oracle OLAP服务器、微软OLAP服务器和Mondrian多维分析引擎提供符合XMLA标准的统一访问。更多还原

【Abstract】 With the development of the database technology and the wide application of the database system,there have been a great deal of history data accumulated in all kind of database system. Online Analysis Processing(short for OLAP) which can find the regulation and refine the useful information from those existing data has been used as the base of enterprise decision-making,and it also emphasized on taking useful information by data analysis based on the DW (Data Warehouse).From another aspect,OLAP,which is an advanced DSS tool,has been more and more popularly used to make deep investigation of business data,find the business tendency and abnomrality,get important details and make better understanding of rules of their business behaviors.However,a majority of OLAP manufacturers have native data access standards instead of uniformed data access standards,therefore, there is a great challenge to the unified access to the data warehouse.In order to achieve the target of uniformly access to heterogeneous data warehouse even realize the integration with them, a majority of OLAP tool manufactures take great efforts to develop the unified OLAP API interface which can access all OLAP server.The necessary to develop unified OLAP API standard which can acesses to all OLAP server can be analyzed from three respects:(1)from the respect of the data warehouse developer, the single tool is rarely able to fully meet the changing needs of the users,and it is also difficult to integrate with various product.(2)from the respect of the data warehouse user, in face of a great deal of data,it can not easily find the really needed information,and it is beyond challenge to completely and correctly express that information.(3)from the respect of the data warehousing vendors,metadata integration is too costly because of lacking of the standard format of the share information.This paper,in the first place.briefly introduces the research status home and abroad along with the related theory of data warehouse,OLAP technology. And then the metadata model of Microsoft Analysis Services.Modrian and Oracle OLAP Server come to understand in order to realize the implementation of the uniformly access to metadata from all kinds of data warehouse.Three standards of access to data warehouse as well as MDX query language is extremely studied.and especially, the XMLA standard and MDX query language is analysised in detail.due to the Oracle OLAP Server does not support the XMLA standard and MDX query language.To achive the target that the realization of uniformly access to data warehouse in accordance with XMLA standard, the mapping between Oracle OLAP metadata model and XMLA data model is firtly implemented.and then the MDX parser which can parse the MDX language based on MDX character must be developed,the overall system architecture is designed in accordance with the above theoretical knowledge.the architecture takes adventage of Brower/Server three-tier structure,front-end presentation layer makes use of the open source JPivot product.which is a custom JSP tag libraries and covers all OLAP functions and operations, especially supporting for accessing the data source via XMLA standard;the control layer takes adventage of the developed engine in accordance with the XMLA standard(short for XMLA Engine);the data layer takes adventage of an Oracle11g database. In the second place, the design and implementation of the Oracle OLAP so as to provide the data source for XMLA Engine.Finally, in order to the realization of the system and test the system, the front-end JPivot can be uniformly access to Oracle OLAP, Microsoft Analysis Services and Mondrian in accordance with the XMLA standard through the XMLA Engine, thus, achieving the target.更多还原