Application Of ODS In The Insurance Enterprise System Integration

This paper has a foothold on the hotspot of Data Integration Study, at the condition of less theories and techniques about Operational Data Store, we will carry on the detailed contrasting research in the related theories first, accounting its current development and combining theories with practice, then give one system architecture of ODS towards the insurance enterprise system based on the related theories; In the implementing step, many further and mature techniques would be adopted to ensure the usability and expansibility of the system. All of this work will be going to enrich the theories and technique of ODS, make ODS to bridge DB and DW, rather than the replacer of DW or DB.In the aspect of theory, this paper would compare ODS with DB and DW in detail, highlight the advantages of ODS, and gather all the currently achievement completely in home and abroad, basing on which, a three-layer architecture would be designed. The architecture is suitable to the insurance enterprise, and made up of DB-ODS-DW. At last, its main functions will be particularly described.In the realization step, this paper would analyze the actual specific need of the enterprise, and complete the building work of the whole ODS system, mainly coding. All of the work could be divided into three parts: the first part is ODS-OLTP, which means ODS can support the on-line transaction process, which decided by two characteristics of ODS — "volatile (including the updated)" and "current-valued", the system architecture follows the principle of J2EE, and in the application designing, it refers to the method of Service-Oriented Analysis and Design, in which the developing is drived by service and business-flow, so that the system can possess of best performance and expansibility; in the second part, Rules of Data Integration are necessary for the data existing in many different structures, the rules can picture the whole process of Uniting and Merging, specially for dealing with the same andincomplete data from many legacy systems, with the rules the enterprise-unified information view can be attained; The last part will introduce ODS-ETL, which means the data's migration process in ODS, in this process the data in the distributed legacy systems would be transferred into the ODS system, the whole process is similar to the ETL process in Data Warehouse, in the paper both data-cleaning step and data-transforming step will be explained in detail, also including the related principles. Finally, some work would be done for the combination between ODS-OLTP and ODS-ETL to make the system to be an organic one.