This paper presents an approach to support the design of ‘dynamic’ databases where large amounts of data are updated frequently. The basis of our approach is formed by schemata of an extended entity-relationship model which are constructed in a modular way and can then be enriched by behavior specifications, functions, and view definitions. The goal is to derive the retrieval and update semantics from a given data schema as far as possible. The components of the database design toolbox RADD can be used to derive and analyze applications of the resulting information system, such that alternative EER representations with better performance properties can be automatically inferred, and visualized to the database designer. In this way, bottlenecks of the modeled system can be omitted – using the conceptual view to the database, and a logical formalism for requirements specification and database tuning.