Databases Reference
In-Depth Information
through its territories, distribution networks, etc. In any
case, since these contexts can change over time, you must
not impose an update of semantic models to take into
account new ones or delete others.
9.3.4.2.
Temporal scalability
At a certain place in the model, the data analyst can
provide for meta-data, which allows a list of data to be
declared in the form of an information structure of the type:
{name of the data, type of data
7
, value of the data, period of
validity, documentation of the data}. For the MDM system,
this meta-data is master data like any other. Consequently,
the user, if they have the authorization, can declare new
data at this place in the model.
For example, for the configuration of products, the data
analyst provides for “additional characteristic” meta-data
which authorizes the addition of information unknown at the
time of modeling. Since this new data can also be attached to
a period of validity, we therefore dispose of a means of
extending the model over time. In our example, we can
indicate that certain characteristics will only be available as
from a date in the future.
This approach, by using meta-data, remains simple and
might turn out to be insufficient. Indeed, the added data is
elementary. It is not possible to declare new complex
information structures. In the same way, it is not possible to
declare rules for the validation of this data. Other
constraints can appear, in particular the management of
rights of access to new data declared in this dynamic way.
Finally, this usage must be reserved for specific cases of
extension of models. Indeed, the data declared are no longer
documented by the model itself, which reduces its quality
7
. String, Integer, Date, etc.
