This article originally appeared on the
One of the things I have been thinking about is how we can balance the long-term strategic need to
properly architect an enterprise data warehouse (EDW), with the ability to respond quickly to the
short-term expectations of the business users, and I wanted to share with you – and get your
thoughts on – what I have come up with.
As you have taught me, the successful data warehouse is built iteratively and in small, fast increments.
The approach is to start with a "thin slice" of the enterprise logical data model (LDM) and introduce that into the EDW. This thin slice can either be "vertical" comprising the major entities from one subject area, or "horizontal" across different subject areas that when brought together and integrated, multiply the business value beyond that of its constituents.
As each slice is introduced into the EDW, the data marts that represent the most effective and actionable subset of data, can be built off that slice, thus demonstrating business value to the end-user community fairly quickly, without waiting for the full-blown EDW to be built.