Research On Historical Data Archiving And Reconstructing Strategy In Data Warehouse

A large amount of particular level data in the data warehouse system is the basis and the important operation objects to the on-line analysis management and the data mining application. In order to guarantee the efficiency and quality, the large amount of obsolete particular level data should be filed as history data.It is discovered that existent filing methods are incomplete through the research for the history data files and reconfiguration methods based on the present XML using a file as a unit in data warehouse. In order to solve the problem, a history data filing method based XML using a record as a unit is proposed. Changing pigeonholing unit, document structure and the relationship between the filed document and XML document are to solve the problem the original methods met. Changing document Ostructure makes XML documents not increase the storage space of the identical information in each edition, and only increase those of changing information and the symbol information. At the same time, after all different editions of the same file operates pigeonholing, they correspond to one XML document. Thus it is able to avoid the space waste to a certain degree caused by that the data change amount is not big. So the data overlapped phenomenon in the each file is eliminated, improves efficiency of the storage space and improves the performance of the retrieval function. However, this method also exits the drawback that is not good at the continuity. The paper proves that archiving improvement strategy is more significant than the original strategy in saving storage space through powerful proof of the corresponding experimental data.In order to have better effective retrieval and improve the efficiency of the storage space in pigeonholing data, a method that archival history data is filed using a record as a unit to realize the classification administration is proposed. Namely according to reuse state of the filed history data, history data in a low frequency are transferred to storage equipment at a lower frequency.