Databases Reference
In-Depth Information
5.5. A
Model-driven
MDM system
In the first chapter of this topic, we explained the
difference between a “transactional data repository” and a
“semantic data repository”:
-
the first is based on a data model that does not accept
validation rules. It corresponds to the classic relational
database, similar to the maturity of a static MDM system;
-
the second is based on a rich data model which
governs the behavior of the data repository thanks to the
validation rules available in the model. This corresponds to a
semantic MDM system, which can also be described, using
its technical label, as a Model-driven MDM system.
Model-driven MDM completes the normal functionality of
RDBMS (Relational Database Management System) with
two notable additions that we will now describe: the
management of data variants and a hiding mechanism for
join tables. Even if this is somewhat technically advanced, it
must be understood by business users as its implementation
is key to the success of an MDM system process.
5.5.1.
Variants
The same data model can contain different characteristics
depending on variations linked to the context in which it is
used. For example, a data element for a business object may
be mandatory according to whether it is used by head office,
a subsidary company, a partner, etc.
Rather than fix the information that specifies the
mandatory parts of the data element, Model-driven MDM
treats it as master data whose value (mandatory or optional)
