Improving Availability for Stateful Applications in Kubernetes
summary
Over 40% of clusters run stateful apps and Kubernetes now supports 15 in-tree drivers and 35+ CSI drivers; storage choices matter. Demo uses six nodes across three zones with four statefulsets (local disk, single-zone cloud disk, regional PD, and NFS) and shows local volumes fail on node loss, single-zone fails on zone loss, while regional and NFS survive. Recovery time includes failure detection and volume detach/attach; features like volume topology, pod anti‑affinity and node auto‑repair (10m) reduce downtime.
AI-written summary · about
chapters (7)
- 0:00Introduction and agenda
- 1:53Storage options overview
- 8:53Architecting for HA
- 16:02Failure scenarios demo
- 28:04Managing pod downtime
- 33:30Summary and conclusion
- 36:15Q&A session
related talks
29mStateful Workloads in Kubernetes: A Deep Dive
33mNot so graceful: K8s node failure and stateful workloads
47mPods & Pages: Databases in Kubernetes
45mKubernetes Resiliency
35mStateful Apps On Kubernetes - Bring Them On!
10m⚡ Lightning Talk: Navigating the Complexities of Stateful Workloads in Kubernetes
30mIs Your Kubernetes Disaster Recovery Actually Ready?
37mYou Have Stateful Apps - What if Kubernetes Would Also Run Your Storage?