Speaker
Description
Relational databases are widely used across research disciplines to manage complex, structured, and evolving data. Yet their sustainable operation remains a largely unresolved challenge in research data management. When projects end, databases are often reduced to static exports or abandoned together with the infrastructure required to operate them. While conventional repositories can preserve database dumps as files, this does not preserve the database as a functional, queryable research object.
DBRepo as a Service (DBRepo aaS) addresses this cross-cutting infrastructure gap. Funded by the German Research Foundation (DFG) and coordinated by the University and State Library Darmstadt, the project is developing a low-threshold, sustainable, and FAIR-oriented service for hosting heterogeneous relational research databases. It builds on the open-source DBRepo software originally developed at TU Wien and aims to preserve databases as functional resources that remain accessible, queryable, and reusable beyond the lifetime of individual research projects.
DBRepo combines database operation with core research data management functionality. Databases and their contents can be described with metadata, queried through graphical and programmatic interfaces, and made persistently identifiable. Version-specific persistent identifiers enable reproducible citation of database states and query results, while semantic descriptions support interoperability and reuse.
A particular potential lies in combining DBRepo aaS with other research infrastructure services. Through its Python library, data stored in DBRepo can be accessed programmatically and loaded directly into environments such as Jupyter Notebooks, including as Pandas DataFrames. This creates a natural synergy with Jupyter4NFDI: DBRepo aaS can provide persistent, versioned, and citable relational data, while Jupyter4NFDI provides an environment for interactive analysis and reproducible computational workflows. Such complementary services could enable researchers to move seamlessly from managed research data to analysis without sacrificing provenance and citability.
DBRepo aaS further develops the platform through richer semantic annotation, AI-assisted natural-language querying, hosting options for database-specific presentation layers, and SIARD-based export workflows for long-term preservation.
The project therefore goes beyond providing another repository. It explores a sustainable service model for a class of research data that currently falls between conventional repository infrastructures and project-specific database hosting. The poster presents DBRepo aaS as a response to a potentially missing cross-cutting service need and invites the NFDI community to discuss integration scenarios with services such as Jupyter4NFDI, interoperability, governance and operating models, and pathways towards broader adoption within a federated research data infrastructure.
| background | external initiative |
|---|