Question :
I want to create and implement a Data Vault Model. In the presentation layer, I want to create a View based on a star schema. Now I would like to know if there is a certain order in creating the models in practice?
(Is is a commonplace to create first a Data Vault model and then the star schema, or vice versa?) I noticed that I think very “dimensional” when start to make a concept of the requirements.
Answer :
Seeing as you are asking about models, I assume you are in the requirements and design phase. If you know what your target dimensional model looks like then you can start there, assuming this is the result of business analysis of the SME requirements.
This can inform the scope of your DV modelling. The combination of these two perspectives will inform your business concept model which will inform you HUB/LINK decisions. Do follow the DV2.0 methodology and principles. Once you’ve built your DV schema, then revisit the dimensional model as it may need refactoring with the newfound clarity.
Data Vault is an architecture and methodology to store data historically, sometimes holding hundreds of entities which can, but don’t need to be interconnected.
Star Schema is an architecture to provide fast querying performance for specific questions, usually holding a few entities which are interconnected.
Without getting any more specific in both methods, I think you might have misunderstood Data Vault. It might be wise to inform yourself more about Data Vault, especially where it might lie in a typical Data Warehouse. Since this is not concerning the original question, I will not go into detail.
I have seen it like this:
- Data Vault is being used as a core storage system. All the data at all the time. No transformation is being made.
- Star Schema is build upon Data Vault Raw / Business Vault. Can be views, can be persistent tables. All calculations / transformations will be done here.