Page 66 -

A Company and its Data 27

relationships and their validation rules. Any intrusion into
silos is still limited as the MDM system can be the object of
an asynchronous technical integration with other databases
(see Chapter 12). This rich data model takes full advantage
of time, context and version management.

It is essential to deal with variants of the same data
model, depending on its use contexts. For instance, the data
cardinality that link two business objects are handled as
master data, which allows different values depending on who
is using the data. We call this level of data “total quality”. It
is necessary to employ a Model-driven MDM system so that
the solution is directly aligned on the semantic model.

1.5. Different types of data repositories

Companies did not wait for the appearance of MDM to
worry about the management of their reference and master
data repositories. These data repositories have always
existed. However, the manner in which they have been
treated has been very variable: spreadsheets, direct IT
access to database tables, tools used under the responsibility
of IT, business software tools which are heterogenous
depending on business departments, etc. Among these
solutions, it is important to distinguish the three that can be
classified as software packages and which are currently used
in companies: Customer Data Integration (CDI), Product
Information Management (PIM) or Product Life
Management (PLM), and a repository for organization
structures based on the LDAP (Light Directory Access
Protocol) standard.

Before explaining the differences between these data
repositories and an MDM system, it is important to first take
into account a technical notion that are unfamiliar to the
management or business world: the difference between
transactional repositories and semantic repositories.