IT once started with the automation of procedural registration processes. The same thing that people were doing until then, using paper and pencil, was copied into a computer. The resulting (information) systems looked like a network of automated procedures.

In this solution, administrated data were kept in many places, in many data files. The problem of inconsistency between these data became very urgent. A solution was found by focussing on the data, and store them only once and in a central place: the databank. With this, an unambiguous administration was guaranteed.
Parts of the former automated procedures still were connected to the centralised data adminstration in their role of input- and output-processes.

Information systems grew bigger and bigger, more input- and output-processes came to surround them. The data all seemed to 'clot' together into one non-flexible mass. And soon the need arose to organise the data more effectively.
Also the amount of input- and output-processes grew, resulting in the need for control of this all. That's why new applications, such as workflow management, came in. Their solutions however could not be easily combined with the existing data model.

A solution was found in an approach, in which organization dependent elements were separated from organization independent elements.
In the centre of the information system one could find an architecture of functional (logical) clusters supporting the administration of 'domain' data, which are organisation independent.
Surrounding them an organisation dependent information system architecture was desinged facing the outside world.

Now an effective architecture for flexible information systems was found. But this logical architecture still had to be elaborated into a design. The question is, how this can be done best:

In fact, going on with the design of an overall data model is a step back into history. The second option might be a much better one, leading to a new approach in information systems design......