PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
February 5, 20260 citations

Latest Developments of the PUNCH4NFDI Compute and Storage Infrastructures

View Full Paper
BRBenoit RolandMGManuel GiffelsAKA. Khalatyan

Key Points

  • The aim is to create effective data management solutions for various physics communities within the National Research Data Infrastructure.
  • Developed federated compute and storage infrastructures—Compute4PUNCH and Storage4PUNCH.
  • Managed compute resources with overlay batch systems and meta-schedulers like TARDIS and COBalD.
  • Standardized access using token-based authentication and automated access token refresh.
  • Integrated Compute4PUNCH as a compute backend for the REANA analysis platform.
  • Enhanced resource optimization through real-time demand matching.
  • Increased usability and scalability of virtualized software environments.
  • Streamlined access management due to automated token refresh mechanisms.

Abstract

The PUNCH4NFDI consortium, funded by the German Research Foundation for an initial period of five years, gathers various physics communities - particle, astro-, astroparticle, hadron and nuclear physics - from different institutions embedded in the National Research Data Infrastructure initiative. The goal of PUNCH4NFDI is the establishment of FAIR data management solutions for the participating communities. The federated compute and storage infrastructures made available to the consortium, Compute4PUNCH and Storage4PUNCH, comprise a variety of heterogeneous compute and storage systems. The compute resources are managed by an overlay batch system and COBalD/TARDIS meta-schedulers. The TARDIS resource manager is responsible for the provisioning of resources and their integration in the overlay batch system based on HTCondor, while the COBalD resource balancer optimises the resource utilization by matching the actual demand for a given type of resources. The access to the resources is standardised using a token-based authentication and authorization infrastructure. The refreshment of short-lived access tokens is automated using the HTCondor Credential Manager and the MyToken service. Login nodes define single entry points to the federation, while the use of containers and the CERN Virtual Machine File System ensures a scalable provisioning of virtualized software environments. The latest developments are presented, including the access tokens management and the integration of Compute4PUNCH as a compute backend of the REANA analysis platform.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Roland et al. (2025) studied this question.

synapsesocial.com/papers/6984345ff1d9ada3c1fb25f8https://doi.org/10.1051/epjconf/202533701264/pdf
Ask AI
Helpful
Bookmark
Share
View Full Paper