Databases Reference
In-Depth Information
8.3.3.
Business object domains
The last stage of the Enterprise Data Architecture,
business object domains, regroups data categories and
consequently semantic classes. On average, it is enough to
have less than about ten business object domains to
represent the entire company.
For instance, we can identify a “
Reality”
domain that
brings together business objects from the real world, such as
client, third parties, club, association, address, country,
vehicle, etc., themselves grouped in “geography”, “intangible
asset” and “person” data categories.
8.3.4.
Data repository architecture
Without the Enterprise Data Architecture that we are
describing, it would be difficult to counter the data
duplication risks that could appear as early on as in data
modeling. For instance, when an “employee” is modeled, how
do you ensure that you do not model certain subjects that are
attached to him/her? If the “geography” data category is
properly identified and isolated from the rest of the models,
it can be reused as a semantic concept to link employees with
their addresses. On the other hand, if the semantic modeling
of the address is diluted in several other models, such as
those that represent real estate assets or third parties, the
reusing of the address model becomes more hazardous.
The damage of data duplication in the semantic modeling
are considerable, going as far as generating silo data
repositories, which we are looking to avoid at all cost in the
MDM approach.
These components (business objects, data categories,
domains) structure the administration of reference and
master data. The MDM tool presents users with the domains
list, followed by the data categories within each of these
