Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»


The star join then has its rightful place as a foundation for data mart design. Figure 3.55 shows how the star join and the data model fit as foundations for data mart DSS design. The star join applies as a design foundation to the very large entities that will exist in the data mart. The data model applies as a design foundation to the nonvoluminous entities found in the data mart.


dimension    fact    dimension


order

Figure 3.55 Classical data modeling applies to the dimension tables (i.e., the nonpopulous entities) and star join design applies to the fact tables (i.e., the populous entities).


order

Figure 3.53 A simple star join in which the entity ORDER is populated with many occurrences and other entities are prejoined with the data.



One of the issues of data warehouses and data marts is how data gets from the data warehouse to the data mart. Data in the data warehouse is very granular. Data in the data mart is very compact and summarized. Periodically data must be moved from the data warehouse to the data mart. This movement of data from the data warehouse to the data mart is analogous to the movement of data into the data warehouse from the operational legacy environment.



Data Marts: A Substitute for a Data Warehouse?


There is an argument in the IT community that says that a data warehouse is expensive and troublesome to build. Indeed, a data warehouse requires resources in the best of cases. But building a data warehouse is absolutely worth the effort. The argument for not building a data warehouse usually leads to building something short of a data warehouse, usually a data mart. The premise is that you can get a lot out of a data mart without the high cost and investment for a data warehouse.

Скачать в pdf «Building the Data Warehouse»