An alternative data warehouse structure for performing updates
Joseph H. Hanson, Mary Jane Willshire · 1996
One key area involved in implementing the data warehouse is the transfer of information from the SDBs into the data warehouse and making it available to the user. Once the information is made available to the data warehouse, it must be incorporated into the overall data warehouse scheme. The difficulty is that the SDBs, being local in nature, are changing at a very rapid rate. In looking at the overall intent of the data warehouse as a repository for vast amounts of diverse information, these changes, made to the SDBs, can be critical to the correctness of the query responses provided by the warehouse. The problem could be solved by extracting all of the pertinent information and changes from the SDBs and immediately incorporating it into the warehouse. The pitfall to this type of solution is that the data warehouse will have very large amounts of information stored in its tables, possibly up to terabytes, and the incorporation and re-calculation of the views for this size database would be extremely expensive in terms of both time and computing power. This thesis presents an alterative to the state of the art data warehouse environment by providing an auxiliary structure to accept these updates and store them until the data warehouse has available cycle time to incorporate the new data and update the necessary views. By performing simulations against both the state of the art and alternative structures, this work proves that there is a significant performance improvement in user query response time by incorporating the auxiliary structure.