The Data Vault Handbook - Concepts and Applications

— With Tracking Changes: Normally, changes to reference data need to be tracked. In this case we would model it using a Hub and one or more Satellites. The Hub stores a unique list of reference codes (e.g., ISO codes), and since these codes are of a fixed-length, we can opt to store them in their readable form rather than hashing them. The descriptive data, which may change over time, is stored in Satellites. This structure allows for tracking the full history of changes, and Satellite splits can be employed to manage, for instance, when multiple source systems are involved. The benefit of this historized approach is the flexibility it provides. You can query the reference data to obtain the latest valid record, retrieve the entire historical record for any key, or define a specific level of granularity, such as using a snapshot for analysis.

REFERENCE DATA EXAMPLE

In a hospital, departments such as Cardiology, Neurology, and Emergency Room are identified by unique codes. This reference data is used to ensure consistent reporting and data entry across different systems. For example, the department code “CARD” represents Cardiology, while “NEUR” rep- resents Neurology. We will present an example of a simple Reference Table without history. The historized version of this table would follow a similar structure but be de- composed into a Hub for the business keys and a Satellite for the descriptive attributes, allowing for historical tracking of changes.