- Since we don't have version 5.0 installed yet at US WEST and I've seen
it very little, some of my ideas may be supported by version 5.0. I'd like
to see the ability to better cross-reference data and process. A CRUD matrix
would document how processes act upon data. Creating a matrix in a tool
other than PowerDesigner would not allow consistency checking to determine
that the entity or process actually exists. The CRUD matrix is an essential
element within some methodolgies used to determine the scope of business
and design areas which would assist in determining how to check out objects
from the enterprise model to the teams.
Another method of relating processes and entities involves modelling
the lifecycle of entities that have major importance to the business or
that have multiple lifecycles. Again, including it in the PowerDesigner product
will force consistency checking.
- I would like to see the ability to create a process-view of the data.
Again, this would be to include only the entities and attributes that are
inputs, outputs or required elements for processing within that process.
This is not a submodel in the manner that PowerDesigner is using them for.
(It appears the submodel is mostly to control the update of objects.) Adding
or changing objects within a view would update the model used to create
the view. (Submodels in PowerDesigner are similar to encyclopedias within
Sterling KEY ADW.) These process information views are also important when
reviewing the models with the business people. You only review the process
information view that relates to each process they are concerned with.
They may not care about attributes that other parts of the business care
about and it takes too long to try and focus only on their piece if they
see it all. It also takes too long to hide or mask information not relevant
to them.
- I'd like to see the ability to do process dependency modelling including
the ability to show how time and events affect the process. You should
be able to model triggers, pre- and post-conditions for each process. These
also relate back to data because pre- and post-conditions usually relate
to the entity lifecycle.
- The ProcessAnalyst should be expanded to include capturing logic and
conditional statements that could be translated to code later. This also
would force determining consistency between the data and process pieces
of the software. You should be able to relate to the data objects identified
and defined in the enterprise model. This is where all the pieces come
together. The process views of the data and the logic input to support
elementary processes are used to create modules of code. (A true enterprise
model from an IE perspective is one that contains data, processes, and
the interaction of the two.)