Data warehouse systems integrate information from several operational systems and store this information in a data storage (the data warehouse, DWH) that is targeted for decision support. In the context of the SIRIUS project, we investigate the data warehouse refresh process with the main focus on

• the applicability of our approach in data warehouse environments consisting of a wide variety of heterogeneous operational sources (various database systems, flat files, etc.),
• providing facilities in order to refresh a DWH independently of how warehouse data are persistently stored, i.e., we make no assumptions about the database system used for the DWH and about modeling concepts,
• the support of incremental warehouse refresh. Data warehouse volumes can reach hundreds of gigabytes or even several terabytes. Hence, avoiding a complete reload of DWH data is a crucial issue and can reduce the refreshment time of DWH data significantly,
• providing mechanisms regarding the special nature of the warehouse data like derived and historical data,
• the support of the automatic execution of the whole refresh process based on user-defined specifications (e.g., about the start of the refresh process),
• facilitating the adaptation of new refresh techniques which might, e.g., become necessary because of the extension of a DWH by new sources or new kinds of data from the same sources.

A. Vavouras, S. Gatziu, K. Dittrich:
The SIRIUS Approach for Refreshing Data Warehouses Incrementally.
8. Fachtagung BTWí99 (Datenbanksysteme in Büro, Technik und Wissenschaft, Freiburg, March 1999