Systematic Modeling Processes are applied by using a step-by-step method to move from data scope to database.
A data model is constructed at each step, utilizing either forward or reverse engineering techniques as needed. The current practice merges analysis and design processes for logical data modeling. With Total Data Modeling, the analysis and design yield their own models separately.
One starts with a data scope model followed by a conceptual data model. Then a data requirements model is constructed from business analysis.
Next, a logical data design model is created either together with a finished data requirement model or separately. Before one gets into the physical data model the data requirements model and the logical data design model are reconciled. Any discrepancies between the two are addressed.
Within the scope indicated by the data scope model, the database is generated from the physical data model after the latter has been enhanced for maintainability and performance.
For reverse engineering, a raw physical data model is then created from a database. The raw model is changed into a rationalized physical data model by separating the requirements-related components from the design-related components. Data mining and semantic modeling are sometimes necessary if the database lacks definitions and constraints specification. This is required in order to infer missing keys and missing relationships. Then, the reverse engineering is completed by following the regular data scope to logical data modeling processes.
In the final step, the logical data model and the rationalized physical data model are reconciled.