Databases Reference
In-Depth Information
8.1.2.1.
Bottom-up
The use of a bottom-up approach is interesting in order to
progressively validate the different model versions, built in
stages. Due to the multiple and often short iterations
brought by this lifecycle, the MDM system must be capable
of rapidly exploiting the data models in order to
automatically generate the necessary user interface for data
governance. This MDM, based on models, corresponds to the
Model-driven MDM (see Chapter 5).
The iterative approach means managing repeating cycles
of data migration, from one data model version to another.
The successive versions can impact the integration layer that
synchronizes the MDM system with the rest of the IT
system.
The lack of a global approach makes stable modeling
(which sufficiently integrates a sustainable architecture
frame early enough) more difficult. Consequently, the
recommended use of the
bottom-up
approach is as follows:
-
the iterative cycle is used with a view to prototyping
the model, without a final loading of data. As soon as the
iterations provide a stable version of the model, the data
loading is carried out definitively. From then on, data model
version management is compulsory and is accompanied by
data migration procedures between versions;
-
from the first iterations, the data model takes into
account best practices that facilitate modifications in upward
compatibility, which limit the burden of data migration from
one model version to another, e.g. the use of complex types
rather than simple ones, transversal domain-gathering
reusable data types, dynamic attributes, etc.
The use of a single iterative approach is not
recommended. It is necessary to benefit from the capacity to
demonstrate and validate data models but it must be
