Generally the industry recognizes three distinct categories of data in a BI environment:
- Transaction data
- Master data
-
Meta data
These different type are treated very differently by most data warehouse systems, in particular the ones that are built from fact tables and dimension tables. Close inspection learns that the distinction is not always clear, and depends on the analysis needs of the organisation. The data warehouse can be (and should be) oblivious of the differences between transaction, master and meta data.
This notion calls for a very generic and flexible data modeling technique. One that is still simple enough to quickly comprehend and scalable enough to be used in any data warehouse environment , from very small scale to multiple-petabytes. We believe such a modeling technique is the Data Vault, as designed and described by Dan Linstedt. Visit our dedicated sister-company www.qosqo.nl for details.