Question

I would like to know the best approach to create a Common Information Model. Just to be clear, I’ve also heard it referred to as a canonical information model, semantic information model, and master data model – As far as I can tell, they are all referring to the same concept.

I’ve heard in the past that a combined “top-down” and “bottom-up” approach is best. This has the advantage of incorporating “Ivory tower” architects and developers – The work will meet somewhere in the middle and usually be both logical and practical. However, this involves bringing in a lot of people with different skill sets.

I’ve also seen a couple of references to the Distributed Management Task Force, but I can’t glean much on best practices in terms of CIM development.

This is something I’m quite interested in getting some feedback on since having a strong CIM is a prerequisite to SOA.

Update

I’ve heard another strategy goes along with overall SOA implementation: Get the business involved, and seek executive sponsorship. This would be part of the “Top-down” effort.

Answer 1

The canonical data model is a data model that is independent from any application or service. Each application that you need to integrate should transform messages from and to this common format.

You should only integrate data in the model that needs to be shared between the applications. Take a look at the messages that the applications/services send/receive. Don’t try to create a single Corporate Data Model, I have never seen this work. Always there are political issues surrounding applications and ownership of applications.

A practical approach for creating a model:

• Start with creating reference tables (standard domain definitions valuta, countries, marital status, cities etc.)
• If there are more that 2 consumers then standardize the common concepts (customer, address)
• If available use or refer to public standards (ISO, ontologies e.g. Dublin Core)
• Explicitly assign ownership for the model.