One task of Bioinformatics is the implementation of a uniform environment for the homogeneous access to the wide spread molecular biology data. Because of the large amount of databases in this domain and the growing knowledge, a mediator based approach for a database integration is presented. In this context a Framework called MARG-Bench was developed. In a subproject a workable approach for mediator based database integration in the field of molecular biology has been developed. It is based on online retrieved data, integration operators, global data schemas and database adapter. Moreover, the developed methods and algorithms are validated in a prototypical client-server implementation - the BioDataServer.