Database modeling is still a job of an artisan. Due to this approach database schemata evolve by growth without any evolution plan. Finally, they cannot be examined, surveyed, consistently extended or analyzed. Querying and maintenance become very difficult. Instead of “development in the small” we propose an approach towards “ development in the large”. The approach is based on the observation that large database applications often use an implicit structuring into connected components. Components are either star or snowflake schemes. Components may be unwrapped or wrapped, i.e. with associated views. The composition of components is based on associating visible types, e.g., types in views of the components. The paper aims in a development of component engineering.