Information Technology Reference
In-Depth Information
Finally, correspondences can also be very useful in multi-viewpoint spec-
ifications for checking consistency between the different views. Part 3 of the
reference model discusses this problem and explains that, if two viewpoint
specifications S
1
and S
2
are related by a set of correspondences that can be
specified in terms of a model transformationT, the transformation can be
applied to S
1
resulting in a new specicationT(S
1
). This can then be com-
pared directly to S
2
in order to perform different kinds of checks. Examples of
such consistency checks include, for instance, behavioural compatibility [60]
between allegedly equivalent sets of objects, or the detection of contradictory
constraints in one or another of the viewpoint specifications.
15.7 More Integration
We are a long way from a situation where a system can be deployed from
a set of viewpoint specifications at the press of a button. In fact, it is un-
likely that we will ever achieve this Utopian objective. Indeed, there are both
positive and negative aspects of the move towards greater integration.
On the positive side, there is the possibility of making more information
available about the way the organization actually works, and feeding this
information back into the system management and control processes. If some
information is available about actual patterns of usage, it can be fed back to
assist autonomous management mechanisms. We can optimize and balance
the use of resources by establishing new paths within the toolchain.
On the other hand, we need to retain a certain looseness of coupling and
compartmentalization to allow proper management. We do not actually want
changes in the enterprise models to feed through immediately into the running
system. We need to observe review processes before changes are accepted,
and to allow designers to experiment with, and assess the impact of, possible
changes before they are put into effect. One of the requirements that goes
hand-in-hand with the progressive increase in the level of integration is the
need for more powerful management tools for the whole process, allowing for
the exploration of options and the circumscription of activities a designer can
carry out, so that they do not exceed their authority.
