Consistency in data warehouse dimensions

Data warehouses present a powerful framework for storing and analyzing huge amounts of data. In this context analyses focus on data that has been gathered over long periods of time, often between one and five years. A data warehouse can therefore be regarded as a specialized historical database. However not only the data kept in a data warehouse has to be seen in a temporal context, but also the fact that dimension data may undergo changes during such a time period needs to be taken into consideration. In this paper we focus on update operations on dimensions and establish a notion of consistency for guiding such operations. We devise algorithms for executing update operations that can be shown to preserve consistency, and we study their time complexity.