The history of the Fact!

It is such a pity that so many designers and -even worse- architects of data warehouses use only 1 architect design for all their customers. It is not called designing under architecture. There is no “one-size-fits-all” architecture for data warehouses. I can image that people still do so because of simplicity in for example implementation. That way even a non data warehouse consultant can implement a data warehouse. It however most often results in poor solutions or failed programs.

Building by architecture means that the design and method is made up specifically for the requirements and the long term plans of the customer. There is not 1 design always the best for every customer.  An Architect must choose or create a good design specifically for the requirements of the customer

There are many methods and designs in the world of building a data warehouse. 2 of them are very famous; The Kimball and the Inmon methodologies. Especially the Kimball method was adopted in the Netherlands and still is. Be careful that i do not mention Data vault, which is more an approach of modelling the data warehouse layer within an architecture but from itself is not created as an architecture. Even better, Bill Inmon himself has adopted the DV approach within his (Corporate Information Factory) architecture.

The supporters of these methods and approaches have an almost religious believe in the correctness of the method they support. However it is hardly ever taken into account that both approaches have different quality requirements. Neither are a best choice. A combination of both most often gives the correct approach of what a customer wants. Sometimes it is even preferable to place both approaches beside each other, for example if the availability requirements for a part of the information strongly differ from the rest.

And no this is absolutely not a fixed law or approach on itself. A project/designer/architect should never use a architecture/approach from the shelf. It still stands that a design starts with the business requirements and long term strategy of the organisation with information management. This along with a list of quality and non-functional requirements with which a specific design/architecture can be selected or made. The standards for these requirements should be find out with the customer/end-users. However often a customer has not clear what these requirements should be. An architect/information analyst/designer then must get these clear by using interviews, study mission, strategy and vision of the company to determine the optimal standard setting of the data warehouse. This standard setting is highly important and must be closely tuned with the end-users/customer/company. When the standard setting has too many quality requirements the price to build/maintain the data warehouse will become extremely high. On the other side when they are set too low the data warehouse will NOT be useful enough or the results will be very disappointed.