Your own personal supercomputer within 15 minutes or less
summary
Claims a full open‑source HPC stack called Charmed HPC can deploy in 15 minutes using Juju to orchestrate Slurm for scheduling, CephFS for distributed storage, LDAP for identity, and Open OnDemand as the web UI. Demo shows Juju deploying Slurm and a CephFS cluster, clients mounting the filesystem, and Open OnDemand serving a file manager; observability is partially implemented. Roadmap lists HA for Slurm, fuller observability, IAM, hardware autodetection and documentation.
AI-written summary · about
chapters (10)
- 0:00Introduction and context
- 1:56HPC versus AI industry trends
- 4:10Core needs of modern clusters
- 7:47Introduction to Charmed HPC
- 9:47Charmed HPC architecture
- 15:48Using Juju for orchestration
- 20:11Charm integrations and models
- 22:12Demonstrations and features
- 26:17Roadmap and future work
- 28:24Q&A session
related talks
29mEvaluating two years in production with Juju and Charmed OpenStack
26mOpenStack in 0 to 45 seconds. Engineering at the speed of light
10mLightning Talk: Does Cloud Elasticity Pay off for HPC Workloads?
16mA novel Cloud High Performance Computing infrastructure for user
38mYour Own Little HA Cloud at the Edge (or at Home)
21mKeynote: CERN Experiences - Ricardo Rocha & Clenimar Filemon
15mStandalone yet at scale: Ironic take on High Performance Computing
21mBack to the original mission: an open cloud operating system