Brussels / 3 & 4 February 2024


CephFS at CERN in view of Disaster Recovery

CERN operates multiple Ceph clusters for the provisioning of block, object, and file storage for the IT infrastructure of the Organization. CephFS at CERN (~11PB) serves the shared filesystems in HPC use cases and as well as a storage persistency layer in the Kubernetes and OpenShift offerings. In the past year we evaluated the usage of CephFS in the context of our Business Continuity and Disaster Recovery strategy. This talk looks at CephFS at CERN, the feature stability/perfomance impact observed when enabling features like snapshots, and the future outlook. We will also report on the increasing need for backups and on the development efforts of an OpenStack Manila driver to provide users with a self-service restic-based backup solution for CephFS


Abhishek Lekshmanan