Databases Reference
In-Depth Information
raises the auditability of data; business users have greater
ease in querying and governing data flows between systems.
Beyond reference and master data, the exchanges
between systems concern transactional data. The latter must
also benefit from a more rich and stable data model than the
unique pivotal format. The modeling of this data is carried
out, progressively, by also basing itself on a common
foundation brought about by business object domains, data
categories and business objects.
From a physical point of view, the data model is derived in
the form of an XML schema
2
used in the following manner:
− in its entirety, in the integration layer in order to obtain
a “semantic mapping”. The canonical representation ensured
by this schema rationalizes and renders more reliable the
data transformations between systems;
− the subset that is relevant to reference and master data
is injected in the MDM system.
This reinforcement of modeling in the integration level
dramatically changes the way in which the data
transformations are carried out. It results in the concept of
“semantic integration”.
2
. The XML Schema standard accepts extensions, which enables it to take
into consideration all the necessary directives to implement, at a physical
level, rich data models, without losing information. See for example, the
extensions proposed by the Model-driven MDM EBX Platform from
Orchestra Networks. An extension case concerns the dynamic
management of data cardinality minOccurs and maxOccurs, managed like
master data rather than frozen values in the schema. Another case
concerns taking primary and foreign keys into consideration, essential to
express the associations between business objects and their referential
integrity constraints.
