If the various sections and parts of Rational Rose RealTime (RoseRT) models were always contained in a single model file, then the migration process would be straightforward, requiring little architectural considerations other than how big the resulting monolithic model would be after migration to DevOps Model RealTime (since the entire model would need to be loaded into memory all the time).
However, this is not a practical arrangement in RoseRT because many developers reside in different teams and typically contribute to a single model or top-level executable. Consequently, the model must be subdivided into smaller controlled units to allow for sharing and proper control within a source control management system such as Git or ClearCase. How a model is divided and organized is complicated further by the fact that a RoseRT model can share controlled unit packages with other models. This fact raises a number of questions: should these shared packages exist in a multi-project workspace environment such as Eclipse, or should they be shared in their original project? Should they exist in the context of their original model owner? There are many ways to arrange the model architecture; therefore, the Rational Rose RealTime model import wizard allows for many different permutations to accommodate various model architecture approaches.