The FAIR-DI-FAIRmat Colloquium on
An Infrastructure for the Infrastructure
by Hans-Joachim Bungartz, Technical University of Munich
took place on April 7, 2022 online.
Naturally, the Computational Sciences, or Computational Science and Engineering, show a significant need for computational infrastructure – be it a classical HPC infrastructure (simulation and optimization), a data infrastructure (data access and management), or a data analytics infrastructure (artificial intelligence and machine learning). While these differ in the concrete underlying hardware, software, and expertise as well as in the way how they are organized and provided, there are also large similarities. For example, obviously, it is not sufficient to put hardware on the table; however, it also cannot work without hardware being put on the table.
In Germany, the NFDI – National Research Data Infrastructure – has been established to create such an infrastructure for the provision of, access to, and usage as well as processing of research data. While the NFDI is organized in domain-aligned nation-wide consortia, there are a lot of topics that pop up in various consortia. As a consequence, each consortium that is part of the NFDI needs infrastructure itself, to be organized either by itself or in cross-consortia collaborations – an infrastructure within and for the infrastructure. This holds in particular for FAIRmat, where the infrastructure branch forms one of seven areas.
In his talk, Hans-Joachim Bungartz will address a couple of general considerations [aspects of a research data infrastructure (RDI); generic vs. domain-specific; central vs. distributed vs. local vs. federated; embedding into existing other infrastructures] and give an overview of what is planned and what currently happens on the behalf of RDI in the FAIRmat consortium.