Databases Reference
In-Depth Information
1.5.1.
Technical classification
We will see in this section that a transactional repository
is based on a classical database, whereas a semantic
repository uses a rich data model.
1.5.1.1.
Transactional repositories
A transactional repository is based on a transactional
data model, i.e. limited to a simple description of data,
without taking into account validation rules, time, context
and version management, which we have already described.
Most of the time, a relational style model is involved, as
already well established. This model needs to be completed
by other descriptions which enhance the knowledge of the
data, in particular, validation rules. Because the additions
are not integrated into the model, issues arise in terms of
understanding by business users and IT maintenance. In
fact, the additions are too often expressed in technical terms
and hard-coded in software.
The weak expression levels of the data model do not favor
the formalization of knowledge but allow good performance
at a technical level. This repository qualifies as transactional
since it is capable of dealing with big interactions without its
response times worsening. Several thousand transactions
can be executed in a short space of time, without the
repository being a problem. Relational database technology
supplies the technical aspects to handle a large volume of
transactions. However, the rigidity of the repository means
that software developments to each modification of the data
model have to be made. The high value administration
functions that are added, such as time, context and version
management, are costly and not easy to deliver via this
traditional software development. The consequence of this
approach is to recreate silos, since creation of each new data
repository is the same as creating a new database.
