The Research And Application On Slowly Changing Dimensions In Multidimensional Model

Enterprise information construction has accumulated a lot of business data for companies. Therefore, the problems faced by companies today is not how to find data, but how to take full advantage of these existing useful data, from which to explore the potential commercial value in order to promote further development of business. The emergence of data warehouse technology precisely meets the urgent needs of the enterprise.Data warehouse provides an integrated data environment, which can carry out a series of comprehensive operation on company historical data, such as extraction, cleaning, integration and so on, and then load it into databases of data warehouse. The data warehouse is subject oriented. Analysts point of view depending on the angles which they observe the data, in order to verify their business prediction. In data warehouse, multi-dimensional data model is one of data models which is most used. According it business data is divided into fact data and dimension data. The fact data is used to measure the operational indicators; the dimension data indicate the various descriptive data.In multi-dimensional data model, dimension is the angle which analysts analysis the business data. Different perspectives can yield different results. On dimension of data warehouse, its value does not always remain the same. This leads to an update issues of slowly changing dimension in data warehouse. Slowly changing dimension is focus on attributes which change in the frequency and change slowly. When the dimension attributes change, the angles are used to observe fact data change, so the result of data analysis is different. When the dimension attribute values change, its changes should be timely reflected in data warehouse databases, so that the data warehouse can correctly track the data before and after change. The event-based update solution for slowly changing dimensions is response to this issue, which combines three existing strategies: directly coverage of the property value, add dimension record rows and add dimension columns. Meanwhile, it uses the event mechanism to accurately monitor the dimension attribute changes in operational databases. The event-based update model for slowly changing dimension saves storage space trough incremental approach, at the same time, it embodies the advantage of real time operations.