Reliable and Cost-Effective: Open Storage Strategies for Kubernetes
summary
Shows concrete choices and tradeoffs: compares Rook-Ceph (scalable file/block/object), NooBaa (multicloud object), Longhorn (lightweight block for edge/SMB) and OpenEBS (per-pod storage for lowest latency), and explains features used—CSI drivers, operators, dynamic provisioning, snapshots, scaling. Recommends matching workload to type (DB→block, ML→file, backups→object), using replicate three times for critical data or erasure coding for cold data, and warns about single-zone deployments, thin provisioning surprises, and 10TB → 30TB replication math.
AI-written summary · about
chapters (14)
- 0:00Open storage strategies
- 3:19Why storage matters
- 6:33Advantages of open-source
- 9:22Kubernetes storage challenges
- 10:08Selecting storage solutions
- 12:16Matching storage to workloads
- 15:05Understanding storage types in depth
- 18:23Smart storage features
- 19:48Reliability versus cost
- 21:52Storage config mistakes
- 24:18Storage ops best practices
- 26:06Open-source storage solutions
- 26:51Storage implementation best practices
- 28:18Key takeaways
related talks
36mStorage Wars
28mPersistent Storage with Kubernetes in Production - Which Solution and Why?
35mEconomics of using Local Storage Attached to VMs on Cloud Providers [I]
15mThe Case for Benchmarking Control Operations in Cloud Native Storage
19mKeynote: Debunking the Myth: Kubernetes Storage is Hard
25mContainer Attached Storage: Open Source Extensions: Kubernetes + Data
37mBuild a Kubernetes Based Cloud Native Storage Solution From Scratch
36mTips and tricks for Kubernetes storage