Databases Reference
In-Depth Information
12.3. Data synchronization
The quality of the synchronization procedures between an
MDM system and other databases conditions the success of
the MDM approach. This approach must not be mistaken for
integration models that we saw earlier. Indeed, once the
coupling level (weak, tight, loose) has been chosen, it is also
necessary to determine the synchronization level: does each
data modification in the MDM system mean a
synchronization with the rest of the IS or is it necessary to
group several modifications together?
In most cases, companies that begin the MDM approach
wrongly perceive only two levels of synchronization, with:
− batch treatments which bring together significant
volumes of data, with a daily or superior periodicity; or
− each elementary data modification.
This perception is not sufficient enough as the
synchronization of data executed to each elementary
modification leads, generally, to a technical dead end in the
ESB. Too many update events circulate on the
communication bus. The teams in charge of the integration
then attempt to bring together several events, in order to
factorize certain modifications, but without any data
modeling to support this effort.
The ESB in this case is used as a buffer in order to avoid
the systems being too in demand from the elementary data
modifications. These constraints lead IT Architects to favor
the batch function mode and limit real time updates, which
penalizes the effectiveness of the MDM system. It is possible
to correct this situation by profiting from the lifecycles of
business objects.
