Most existing concepts in data warehousing provide a central database system storing gathered raw data and redundantly computed materialized views. While in current system architectures client tools are sending queries to a central data warehouse system and are only used to graphically present the result, the steady rise in power of personal computers and the expansion of network bandwidth makes it possible to store replicated parts of the data warehouse at the client thus saving network bandwidth and utilizing local computing power. Within such a scenario a - potentially mobile - client does not need to be connected to a central server while performing local analyses. Although this scenario seems attractive, several problems arise by introducing such an architecture: For example schema data could be changed or new fact data could be available. This paper is focusing on the first problem and presents ideas on how changed schema data can be detected and efficiently synchronized between client and server exploiting the special needs and requirements of data warehousing.
展开▼
