Canonical Data Model for Data warehouse

Would like to get your views on suitability of canonical data model for data warehouses. Most of my experience is with star schemas where the ETL is tightly coupled with the source and target data structures. I am seeing articles around canonical model that promises to be loosely coupled . One of the articles I read was referring to a design with common minimal attributes in the dimensional model and extended by XML schema for the extended attributes.

I am finding it difficult to convince myself canonical will work for enterprise data warehouse. Any thoughts ?

Good luck.

You typically find such models in applications where data is collected from many different sources, the contents vary widely and there is a need to integrate it. An example would be reference data in the securities/financial industry.

Such models require a software layer to integrate and present the data in more traditional forms. It is not particularly well suited for an end-user accessible DW becuase of this complexity. However, it can be used in a hub arrangement that publishes more traditional views of the data for consumption.

But 99% of the time, a DW is fed from structured, known, well-defined sources. Creating a canonical model of such data is simply not necessary or desireable.