Keynote: Supporting Swiss Academia with Ceph & OpenStack

on YouTube ↗23m

" summary

SWITCH runs an OpenStack IaaS backed by multiple Ceph clusters in production since 2014, scaling from an eight‑node experiment to ~400 servers and 10 petabytes raw on spinning disks with NVMe pools planned. Workloads include microscopy (0.5 TB per mouse brain → ~60 TB/year) plus video and sync services; 30,000 users store 240 TB on SwitchDrive. Ceph proves reliable — never lost data — though hard failures happen, e.g. a kernel bug causing 512-byte TCP fragmentation that only a kernel upgrade fixes.

AI-written summary · about

" chapters (9)

1. 0:00Introduction and keynote setup
2. 1:47Role of SWITCH in academia
3. 5:20Transition to cloud and OpenStack
4. 9:54Ceph adoption and evolution
5. 13:10Ceph troubleshooting stories
6. 16:30Ceph’s versatility and limitations
7. 18:40Switch Drive and Ceph success
8. 20:47Real-world Ceph applications
9. 22:11Future and conclusion