March 05, 2019 - by CSCS

In the SELVEDAS project, CSCS and PSI reinforce their partnership: Thanks to the Swissuniversities P-5 grant of 684.000 CHF, both institutions are now able to start an ambitious project to realise interactive, extreme data scientific workflows across CSCS and PSI. The P-5 grant promotes access, processing and safeguarding of scientific information across Switzerland (more details regarding the current funded projects can be found here >).

Systematically coupling IT infrastructure with academic data centres
Due to ongoing and future progress in accelerator and detector technologies, large-scale experimental facilities such as the Swiss Light Source (SLS) and the free-electron X-ray laser SwissFEL at the Paul Scherrer Institute (PSI) are projecting unprecedented increases in performance. These developments are leading to rapid growth of data generated during experiments.

The gained data will be difficult, if not impossible, to manage with traditional institutional IT infrastructures. One strategy is to systematically couple the IT infrastructure of these large experiments with scalable compute and data systems of large academic data centres such as the Swiss National Supercomputing Centre (CSCS). In addition, Cloud technologies have successfully introduced a service-driven, on-demand model which, once coupled with traditional HPC technologies, can potentially transform service delivery at HPC data centres. These new technologies introduce key features such as interactivity, on-demand provisioning and service federation.

Usable for other communities
The SELVEDAS project, partially funded by the Swissuniversities together with contributions from CSCS and PSI, aims at realizing PSI use cases by deploying CSCS compute- and data-resources during and after an experiment. In providing such a cloud interface to its services, CSCS will not only enable PSI use cases. The resulting sustainable and generic interface will also be usable for any other scientific community relying upon HPC scalable compute and data capability.

The SELVEDAS project is divided into six work packages that address different technological developments such as porting PSI computational-intensive applications to accelerators (GPU), and enabling a RESTful API to access CSCS services to improve productivity of emerging on-demand, extreme data workflows. Previously, PSI and CSCS have collaborated in the PetaByte-Archive project to store PSI experimental data at CSCS. This collaboration will be extended by the SELVEDAS project.

The P-5 grant on scientific information: access, processing and safeguarding, aims to bundle the offering of the different universities in Switzerland for the provision and processing of scientific information. Further information: https://www.swissuniversities.ch/de/organisation/projekte-und-programme/p-5/