Research data management is of the utmost importance in a world where research data is created with an ever increasing amount and rate and with a high variety across all scientific disciplines. This paper especially discusses software engineering challenges stemming from creating a long-living software system. It
aims at providing a reference implementation for a federated research data infrastructure including interconnected individual repositories for communities and an overarching search based on metadata. The challenges involve a high variety of evolving requirements, the management and development of the distributed and federated infrastructure that are based on existing components, the piloting within the use cases, the efficient training of users, and how to enable the future sustainable operation.