Ceph at 20 Years! Still the Best for Modern Storage
summary
Ceph marks 20 years and claims continued primacy by combining a unified block/object/file stack with concrete systems: CRUSH consistent hashing for deterministic placement, placement groups (default sizing that yields ~100 PGs per OSD) to speed recovery from hours to minutes, and BlueStore (replacing filesystem-backed OSDs in ~2016) for direct-device performance. RBD arrives as a 2010 community contribution, big‑bang tests hit ~20,000-disk imports at CERN, Crimson work begins around 2020, RGW adds multi-site S3 and scale-out metadata for AI, and Ceph repeatedly finds real-world bugs (LZ4 corruption 2020, faulty NVMe/HDD behaviors) that get fixed.
AI-written summary · about
chapters (8)
- 0:00Introduction and 20 years of Ceph
- 3:54Brief history of Ceph
- 9:39Why Ceph remains relevant today
- 14:23Why everyone uses Ceph
- 16:48Crush and placement groups explained
- 20:18Ceph’s reliability and debugging hardware
- 25:18Ceph’s speed and Linux of storage
- 27:09Community and open-source impact
related talks
36mKeynote: State of the Cephalopod
56m25 Years of Storage Research and Education: A Retrospective
21mThe Resilient Hard Disk Drive (1984 to 2062)
20mCERN OpenStack User Story
14mFoundational HPC Systems for 2020 and Beyond
49mKeynote Address: Everything You Always Wanted to Know about Storage Analysis...
05mBare Metal to Containers to K8s: How We Got Here
01h 05m20 years gone — what can we learn from Git's history?