Rook: Intro and Deep Dive With Ceph Storage
summary
Rook deploys and manages Ceph inside Kubernetes and now delegates CSI configuration to Ceph CRs — Rook 1.20 uses Ceph CSI CRs — with NVMeoF and RGW account support marked experimental and a two‑node mode requiring fencing (pacemaker). Ceph CSI claims high throughput — 60 PVCs in 30s, PVC <0.5s — and scales to tens of thousands of volumes. Erasure coding examples show EC 4+2 ≈ 50% storage cost vs replica3; Ceph v20 narrows EC performance gaps and v21 promises direct reads, OMAP and pool migration. DR uses snapshot async replication with configurable intervals (example: 5 min snapshots → minutes of RPO) plus RBD mirroring and VRG.
AI-written summary · about
chapters (9)
- 0:00Welcome
- 1:03Rook and Ceph storage
- 5:01Recent Rook and Ceph CSI features
- 7:06Ceph CSI driver deep dive
- 12:28CSI disaster recovery add-ons
- 14:30Ceph CSI operator overview
- 15:34Erasure coding with Rook
- 21:13Disaster recovery strategies
- 27:53Q&A
related talks
27mRook: Intro and Ceph Deep Dive
40mRook: Intro and Deep Dive Wi...
44mContainer Management at Scale
35mCritical Troubleshooting Tools for Rook Clusters
28mA Welcoming Ceph Experience Improvements for First-Time Users or: How to Make Beginners Love Ceph
29mReliable and Cost-Effective: Open Storage Strategies for Kubernetes
40mJuggling Petabytes: Managing Ceph at Scale with Ceph-ansible
08mRapid Design and Effective Operating of a General Purpose Object Storage at RWTH