Nowacki, S.S.NowackiLukas, U. vonU. vonLukas2022-03-092022-03-092003https://publica.fraunhofer.de/handle/publica/343413PDM federation is a powerful concept for handling the distribution of product data over various data sources (PDM, ERP,EL) -- even across company boundaries. Typical problems in this context include the variety of data models, the integration of different access interfaces and efficient implementation of the business logic. Even though partial solutions do exist for all areas, there is no general approach that functions efficiently with very large data sets and which is also easy to implement in a given environment. Our PDM federation concept solves these problems by combining concrete PDM systems into a single, virtual PDM system. This approach conceals the internal complexity of federation and offers a uniform view and access on the data. This virtual PDM layer does not perform replication of product data, but is responsible for distribution of queries and joining the partial results to produce an overall result on the basis of the distribution model. A precondition of this concept is the mapping of the different data models and access interfaces of the individual PDM systems to form a uniform data model and a uniform access interface. The PDTnet Schema has been identified as a promising foundation for a neutral data model. The schema has been jointly developed by important companies in the German automotive industry and successfully used in various integration scenarios. In order to provide an efficient access interface to connect distributed modules at service level, an efficient Java binding for the XML-based PDTnet interface has been developed. Some foundation modules that can be configured to federate arbitrary PDM systems use this. This paper not only presents the technical concept but also initial results of a pilot project carried out at a German OEM.enproduct data managementdata exchangedistributed data managementapplication integration006Efficient and convenient federation of product dataconference paper