Repositories and Computation: Crossover Episode

Research increasingly becomes data-driven, with substantial amounts of information being generated and analyzed to produce new insights and discoveries. Since Virtual Research Environments (VRE) like JupyterHub [1] are becoming more available, the audience that uses these compute resources becomes more heterogeneous too, demanding seamless identity management, datasource integration and high availability from VRE service providers. To make VRE research outputs more FAIR, they can be deposited in dedicated research data repositories. At TU Wien, we provide two different research data repositories [2,3] for publication of research data results, and Jupyter [4] for data processing as part of our VRE.

In order to improve user experience, we are building a library to be used in the Jupyter notebooks that allows researchers to use datasets directly in the VRE seamlessly provided by the repositories in the background, abstracted from the researcher.

Because the library only depends on public APIs, it is not tied to our VRE and can be used in other deployments as well. Further, we aim to keep the design of the library intentionally simple so that it can be extended with support for additional repository types.

[1] https://jupyter.org/hub
[2] https://researchdata.tuwien.ac.at[
3] https://www.ifs.tuwien.ac.at/infrastructures/dbrepo/
[4] https://jupyter.hpc.tuwien.ac.at/