Using OpenStack to reduce HPC service complexity
summary
Describes Cambridge University’s use of OpenStack (Ironic for bare metal and KVM VMs) to manage a shared pool of hardware that can be partitioned into different platforms: large Slurm bare‑metal clusters (e.g., Wilkes‑3), trusted research environments, on‑demand AI and big‑data platforms. Explains combining bare‑metal OpenStack with Cluster API‑provisioned Kubernetes and KubeApps/JupyterHub to deliver both interactive on‑demand services and traditional HPC clusters from the same infrastructure.
AI-written summary · about
related talks
27mKeynote: OpenStack and the Software-Defined Supercomputer
43mSupporting NASA Science with OpenStack
16mA novel Cloud High Performance Computing infrastructure for user
23mOpen Matters: HP & OpenStack
31mDeclarative Underlays: Scaling Purpose-Built Infrastructure(Clusters)...
34mFishfingers and Custard: An OpenStack and K8s Story
13mAn open cloud for researchers, staff and students
21mKeynote: CERN Experiences - Ricardo Rocha & Clenimar Filemon