Kubernetes Stateful Application Disaster Recovery with Zero Data Loss
summary
Shows a Kubernetes disaster‑recovery stack that achieves RPO zero by using an external Ceph stretched cluster (replica 4, crash‑map) accessed via Rook CSI, with multi‑cluster orchestration by Open Cluster Management and automation by RamenDR. Writes synchronously between two data sites (recommended ~10 ms latency), uses an arbiter tolerating ~100 ms, fences with Ceph OSD blacklist, and demos Rocket.Chat preserving every message and a measured ~2 minute RTO for failover.
AI-written summary · about
chapters (10)
- 0:00Introduction to talk
- 1:45Kubernetes storage DR options
- 5:57Metro DR synchronous replication
- 9:24Metro DR architecture overview
- 13:35Ceph stretch mode basics
- 21:47Failover and failback process
- 26:00Ceph stretch mode internals
- 29:12Live demonstration setup
- 33:31Demo failover and RTO
- 34:13Q&A session begins
related talks
35mDisaster Recovery of Stateful Applications in a Multi-Cluster Environment
30mIs Your Kubernetes Disaster Recovery Actually Ready?
37mDisaster Recovery Options running Apache Kafka in Kubernetes
06m⚡ Lightning Talk: Disaster Recovery - How IaCaC and Kubernetes Enables Cost Efficiency and Fast Recovery
15mWhen Clusters Collide: The Misadventures of Disaster Recovery in K8s
35mDeploy Ceph in Kubernetes: Rook Deep Dive
30mWhatever Can Go Wrong, Will Go Wrong – Rook/Ceph and Storage Failures
42mImproving Availability for Stateful Applications in Kubernetes