DataPLANT receives positive feedback from the DFG, but still work ahead
Mon Feb 10 2020
DataPLANT was present at the 3rd NFDI Community Workshop - Services for Research Data Management in Neuroscience hosted at the 10th Februar at Ludwig Maximilians University Munich to exchange with other NFDI consortia and discuss common topics like infrastructure provisioning. The NFDI4Neuro as well as the NFDI4BIMP intend to hand in their proposals in this year’s round of applications. The main topic of this community workshop was the exchange on infrastructure for the NFDI and especially on the providers’ perspective.
The provider’s perspective in the NFDI was elaborated by the BioDATEN science data center. Providers of research IT infrastructures are faced by significant technological changes especially fostered by resource virtualization. Many of the modern services and workflows are operated in an increasingly cloud-like fashion where data and compute moves into centralized, aggregated resources. Such shared resources allow to faster host new projects. The necessary exess capacity is much easier to maintain and justify in centralized resources as the shared overhead to provide is typically much less than in independent systems. Grant providers start to understand the changed technological landscape and to adopt their funding schemes allowing to buy-in into existing resources preferred to establishing single ones per new project. Users are faced by difficult to forecast requirements. It is often impossible for them to define the exactly “right” configuration of a required resource (sizing challenge). These challenges are answered by the IT industry and science driven cloud and aggregated HPC offerings. The aggregation of resources into larger ones can focus the increased efforts for market analysis, system selection, proper procurement and operation of (large scale) IT infrastructures onto few experts. Further on such a strategy would eliminate the contradiction of typical project run times versus the (significant) delay for equipment provisioning and the usual write down time spans of that equipment.
The massive changes in the IT landscape and of the user expectations increase the pressure for re-orientation of university (scientific) computer centers. For many of them cooperation is the chance to significantly widen their scope of IT services. It helps to keep up with the demand by the scientific communities and to offer a relevantly complete service portfolio. Organizationally, it allows for specialization and community focus. When defining future strategies and operation models they might find a new meaning in supporting research data management by providing efficient infrastructures and consultation to the various scientific communities. Further on it offers them the opportunity to participate in infrastructure calls. These developments offer for the researchers the offloading of non-domain specific tasks and services. Suitable governance structures are to be implemented to ensure a persistent relevance of future computer centers through user participation and feedback loops. Close cooperation and consultation (like already done in Freiburg for the bwForCluster NEMO and for the storage infrastructure bwSFS) helps all stakeholders to have suitable, up to date infrastructures tailored to their needs. Such structures are in their infancy for the NFDI, but future NFDI wide coordination should advance this topic.
The financing of IT infrastructures for the various scientific communities is often grant driven (and inherently not sustainable) for sustainable long term services and research data management. The future would see a changed flow of funding from simple project driven and organization centered practice to demand-driven streams to different providers.Large infrastructure initiatives like de.NBI or the NFDI need not only to solve the role of personnel employment (permanent vs. project based) but to define suitable business and operation models compatible with the VAT regulations and the federal and state requirements for cash flows in mixed consortia.