Augment HPC services with private cloud services

High-performance computing (HPC) has become a key tool for many areas in computational science such as fluid dynamics, biophysics, quantum chemistry or material sciences. HPC systems are also increasingly used for analysing extreme-scale data sets and applying machine learning methods. Sweden is among the leaders in the field and partner in numerous large-scale international initiatives.

The number of Swedish academic institutions and industrial companies taking advantage of extreme-scale compute resources is rapidly increasing. HPC systems offer great capabilities but operate in a protected environment, which hinders external access.

This is in contrast to the worldwide increasing need for integrated access infrastructures for optimal resource utilization.

To bridge this gap, this project suggests to augment the national HPC system, Dardel supercomputer at PDC, with a private cloud instance.

This instance will provide flexible, open and elastic access to computational and storage resources.This augmented infrastructure will serve a broad range of scientific areas.

It will address the emerging needs for complex workflows, where simulations and data analytics on HPC systems are only one component.

E.g. smart cities research relies on sensor data, which is collected outside of HPC centres but needs to be processed through large-scale simulations on HPC resources.

Another example is computational drug design where much experimental and model data is accessed through public repositories and typically used as part of complex workflows including HPC simulations.

HPC augmented cloud services will make easier data discovery, access, integration, and reusability following the FAIR principles.

Importantly, cloud portals can enable rapid deployment of computational jobs in bursting mode on compatible infrastructures.