Databases Reference
In-Depth Information
The constraints induced by these integration models on
the business modeling (semantic) and on the organization
modeling (pragmatic) are expressed primarily on two levels:
− first of all, by the reminder of the importance of the
modeling of the business objects' lifecycles (first line of Table
12.1). In particular, a MDM system in weak coupling means
the management of states in order to synchronize, in a
reliable manner, the data with the rest of the information
system. In other words, the integration model in weak
coupling reaffirms the necessity of opting for the “semantic
MDM” maturity level, uniquely capable of taking into
consideration the management of states as master data (see
Chapter 5);
− then, the data approval processes may have to take into
account organizational steps imposed by the technical
integration of the repository. For example, with the weak
coupling MDM system, if the address is not validated by the
MDM system, it is necessary to push a request in the task
list of a CRM actor in order to correct the problem. This
organizational step exists only if the integration model used
is that of an MDM in weak coupling.
12.2. Semantic integration
The three integration models that we have just described
require the use of a software layer in charge of data
exchanges between systems; generally, this is a
communication bus or Enterprise Service Bus (ESB) .To
avoid this bus becoming a connection nightmare, a “point to
point” between systems, it is necessary to implement a pivot
format of data. Unfortunately, most of the time, this format
is specified in an incremental manner, as new data exchange
needs appear. As time goes by, the pivot format piles up data
structure duplications and has no well-documented data
validation rules. Just like functional and technical silos, we
find in the integration layer a stratification that generates
