In this chapter we saw that transformations can be viewed as more than just processes. Because transformations are used in situations where we cannot generate hundred percent complete target models, we need to be able to support changes in both the source and target models. This can only be achieved if we know the relationship between the changing models.
A transformation can be viewed as an instance and thus be able to keep information about transformations between specific models. This allows tools to support changes in both models, while still keeping them consistent. Transformation instances can also hold parameters and/or user choices that were given during the transformation process and make sure that these are stored and reapplied in future transformations as well.