Databases Reference
In-Depth Information
When a company faces a situation where its data assets
are spread out, it is not uncommon, for example, that the
concept of a factory exists in multiple heterogenous
databases. How is it then possible to guarantee that a
product to factory association is correctly situated? Is there
more than one meaning to the factory concept? Which data
model should be used as a reference? Worse still, if the link
between product and factory is handled at the time of
process integration between the two databases, it is highly
likely that the referential integrity constraints would be
hidden in the EAI/ESB software, without proper
documentation. Due to the fact that there are multiple
associations between data and that companies do not
properly master the links between databases located within
their silos, it is understandable that the general quality of
data is in freefall.
The data
squat
is noxious
In a number of systems, quality defects are such that
the same piece of data can have different meaning
depending on how it is used. IT experts call this a
data
squat
; not knowing how or no longer daring to
improve the existing data model, they employ a part
of the existing model and give it a semantic of a
different kind, with coding enabling a distinction to
be made between data, giving an idea of its origin,
and any hidden meaning. For example, a piece of
data relating to a contract can be
squatted
so that it
suddenly matches a type of insurance guarantee.
This type of madness is not unusual. A number of IT
systems hide this toxic mechanism. With time, it
renders the semantics of the data model useless; only
a proper reconstruction of a semantic model can
enable order to be restored in these systems.
