Design and Implementation of an Enterprise Data Warehouse

The reporting and sharing of information has been synonymous with databases as long as there have been systems to host them. Now more than ever, users expect the sharing of information in an immediate, efficient, and secure manner. However, due to the sheer number of databases within the enterprise, getting the data in an effective fashion requires a coordinated effort between the existing systems. There is a very real need today to have a single location for the storage and sharing of data that users can easily utilize to make improved business decisions, rather than trying to traverse the multiple databases that exist today and can do so by using an enterprise data warehouse. The Thesis involves a description of data warehousing techniques, design, expectations, and challenges regarding data cleansing and transforming existing data, as well as other challenges associated with extracting from transactional databases. The Thesis also includes a technical piece discussing database requirements and technologies used to create and refresh the data warehouse. The Thesis discusses how data from databases and other data warehouses could integrate. In addition, there is discussion of specific data marts within the warehouse to satisfy a specific need. Finally, there are explanations for how users will consume the data in the enterprise data warehouse, such as through reporting and other business intelligence. This discussion also includes the topics of system architecture of how data from databases and other data warehouses from different departments could integrate. An Enterprise Data Warehouse prototype developed will show how a pair of different databases undergoes the Extract, Transform and Load (ETL) process and loaded into an actual set of star schemas then makes the reporting easier. Separately, an important piece of this thesis takes an actual example of data and compares the performance between them by running the same queries against separate databases, one transactional and one data warehouse. As the queries expand in difficulty, larger grows the gap between the actual recorded times of running that same query in the different environments