The Design And Implementation Of Metadata Management Model Based On BI-Paas Platform

With the development of information technology, the massive growth of enterprise data and increasingly complex processing logic, traditional BI systems can barely meet the performance needs of the enterprise. Market demands determine the development direction of the BI system. Based on this fact and considering the advantages of BI and PaaS, integrated development of BI-PaaS platform appeares. The BI-PaaS platform can not only decrease the difficulty of the BI system development but also improve the performance of the BI processing. However, every coin has two sides. There are some problems to be solved in the BI-PaaS platform at the same time. Such as that metadata does not have a universal definition while existing BI tool integration to PaaS platform. Because the metadata description of the abilities and components were different from each other, we could not integrate different ability in an unified manner(example:Report, Data-mining, ETL, etc.).From the metadata point of view, BI-PaaS platform mainly suffers from the following problems:1.Different types of clutter in metadata leads to lack of overall metadata classification and format specification,2.Different roles of metadata are not clearly defined,3.The description and configuration information of BI ability are embedded in the code and lack of unified abstract model, which makes it against the integration of new BI service ability. Based on the above problems, this thesis focuses on the design and implementation of metadata management model based on BI-PaaS platform. Through metadata management development method to exert the strength and minimize the influence of weakness. It enables us to understand the composition, conversion, analysis and processment of data in BI-PaaS platform in global view. It also enables us to manage metadata more effectively and improves the usability and data quality.Firstly, this thesis classifies the metadata of the BI-PaaS platform as technical metadata, business metadata and administrative metadata.Secondly, reference to the metadata management methods of other platforms and each function module of BI-PaaS platform is made of, this thesis divides the metadata model levels and builds a metadata model system,especially on the BI metadata model, the workflow metadata model, the organization metadata model and the UI metadata model.Thirdly, this thesis determines the content of the metadata management framework. The metadata management framework here is focused on the metadata of BI ability.In order to prove the metadata model framework and the meatdata management’s correctness and rationality, this thesis analysis different BI engine cases in BI-PaaS platform. Last but not least, this thesis points out its insufficiency and offers some suggestions for the future study.