Responsibility for HPC data storage provisioning

Determine whether the responsibility for providing short-term and archival storage for data produced and used by high-performance computing workloads should lie with HPC centres, universities, or individual research groups, in order to guide planning for next-generation HPC systems.

Background

The report highlights that biomolecular applications on HPC generate substantial data volumes, from tens of gigabytes per day for molecular dynamics to terabytes for cryo-EM datasets. Reliable short-term and archival storage, as well as efficient data transfer mechanisms, are essential for effective use of HPC resources.

Despite the critical nature of storage infrastructure, the authors note a lack of clarity regarding which parties—HPC centres, universities, or individual research groups—should be responsible for provisioning and maintaining storage. They emphasize that resolving this question is necessary before deploying the next generation of HPC platforms.