Comparative Study of Data Warehouse Architectural Design Approaches

The development of Data Warehouse starts with requirements gathering, designing the dimensional model which is further followed by testing and maintenance. It is the central repository where all the historical data is stored and maintained which is further used for the analysis. The most important activity in building of a data warehouse successfully is design phase. This work presents a brief description of different architectural approaches and techniques that address the DW Design problem. Different data models have been proposed for data warehouse design but these approaches are based on their own visual modeling languages UML or ER model, and there is no standard method or model that allows us to model all aspects of a DW. There are four stages of data migration in the proposed model: Data extraction, cleansing and refining data, data transforming, data indexing and loading which is further explained in the paper. Also this work presents the brief description of different approaches and techniques that addresses the DW Architecture. This paper shows the typical data warehouse architecture for different level: one, two, three level classical and novel three level architecture. The proposed architecture exhibits some drawbacks at the point when connected to work over extensive number of heterogeneous data sources.