Page 69 -

30 Enterprise Data Governance

production databases. These databases then follow classic
optimization procedures used in a relational oriented
approach.

Finally, from a technical point of view, a semantic
repository forms an abstraction layer above the classic
systems of database management, such as DB2, Oracle,
SQL-Server, etc.

Consequently, the power and reliability of these databases
8
are always used and taken advantage of by MDM .
1.5.1.3. Model-driven data repositories

An MDM system must be based on a semantic model. It
must be capable of taking into account any rich data model,
whatever the data domain: product configuration,
organization and third party description, financial
classification, charts of accounts, technical parameters, etc.
This MDM system can also qualify as a Model-driven MDM
system, because the data repository is largely governed by
the rich data model.

Without taking into account the semantic repository,
MDM remains a similar solution to a hard-coded software
development based on a rigid and frozen database. The use
of a database schema from the transactional world (a
relational DDL) in order to deal with the data governance
needs of the reference and master data is not relevant. MDM
needs agility and a Model-driven approach based on a rich
database (XML) schema.

Throughout this book, except when clearly mentioned, the
term MDM will mean Model-driven MDM.

8. From a technical point of view, MDM acts in this case like a DAL (Data
Access Layer) or, more precisely, ORM (Object Relational Mapping).