Page 70 -

A Company and its Data 31

1.5.2. Customer Data Integration (CDI)

It is not uncommon for a company’s customer knowledge
to be spread out over a number of databases located in a
number of systems, such as in a Customer Relationship
Management (CRM) database, an Internet database, a
billing system or a marketing database, etc.

To synchronize these databases, companies complete their
technical integration layer with a storage solution in the
form of a transactional hub, called Customer Data
Integration (CDI).

The image of a hub illustrates well how CDI operates: all
the modifications to customer data are sent to the hub which
is responsible for verifying their validity, keeping data in its
database and notifying other systems of the update, often in
real time. The higher and more frequent the quantity of
exchanged customer data between systems is, the more a
CDI looks like a classic database, i.e. a transactional
repository.

Its transactional foundation does not enable CDI to
manage a rich data model. Therefore, a large part of the data
validation rules and referential integrity constraints are
hard-coded in bespoke software outside the CDI. In other
words, the rigidity of the data models leads to a deficiency in
data governance functions. CDI is not driven by the data
model; it is not Model-driven. Consequently, the data
governance functions are frozen at the start and cannot be
easily adapted to the evolutions of the data model. It is a
rigid approach, of a heavily integrated software package
kind.

Even though the use of the semantic data model to
automatically generate the data governance functions is
appealing, Model-driven system cannot always replace CDI.
The highly transactional aspect, especially in the case of CDI