Databases Reference
In-Depth Information
8.1.2.2.
Top-down: Enterprise Data Architecture
Making the effort to achieve global modeling ensures
greater data model stability and a better upward
compatibility of successive versions. Nonetheless, this “top
down” approach requires a good knowledge of the business,
and mastery of the modeling procedures. The objective is to
establish a data architecture landscape map across the
whole of the Information System, before detailed data
modeling. This is an Enterprise Architecure approach
applied to data.
The choice between the iterative lifecycle (bottom-up) and
the Enterprise Architectural (top-down) approach is not
specific to an MDM project. It is present in all modeling
domains: reference and master data, transactional data,
processes, etc. In most cases, to support a sustainable
approach, the design of the rough components of the data
architecture, across the whole Information System, must be
agreed upon. These components organize the foundation of
the data models. Then, in agreement with this foundation,
the detailed modeling of data is done thanks to an iterative
approach. In the end, this reconciliation of bottom-up and
top-down
approaches
results
in
the
following
recommendations:
-
proceed, as soon as the MDM project is launched,
towards a modeling of the highest level architectural
components, i.e. the business object domains and the
identification of business objects within these domains. At
this stage, you are not yet looking to model information that
describes the content of business objects;
-
this level of modeling is sustainable. If the semantic
modeling procedures are mastered, there is no reason to
modify the resulting architecture in depth; it is a solid and
sustainable data foundation. Nonetheless, evolutions remain
possible and follow a maintenance cycle involving a
Enterprise Architecture responsibility across the whole
