Persistent Storage with Kubernetes in Production - Which Solution and Why?
summary
Compares storage by concrete tradeoffs: performance, access pattern, deployment and management. Recommends object store (S3/EFS/minio) for backups and large blobs, shared filesystems for media, block storage for Postgres where deterministic latency matters. Warns cloud volumes (EBS/PersistentDisk) have attach/detach 45s–1hr delays that affect failover. Presents eight principles of cloud native storage and the emerging CSI standard. Demos a software defined system that runs a container per node, aggregates local blocks, and uses synchronous replication to 0–5 replicas with instantaneous failover via etcd-based control.
AI-written summary · about
chapters (10)
- 0:00Introduction and overview
- 2:07Why state is tricky in Kubernetes
- 4:15Comparing storage options framework
- 7:55Kubernetes storage model
- 11:35Choosing storage for WordPress
- 14:28Key questions for storage selection
- 19:31Postgres database storage
- 22:04Software-defined storage demo
- 24:18Recap and CSI introduction
- 26:27Q&A session
related talks
36mPersistent Storage with Kubernetes in Production - Which Solution and Why?
42mDistributed Storage in the Cloud
49mDevOps: Kubernetes Storage - Myths, Facts and Tips
29mReliable and Cost-Effective: Open Storage Strategies for Kubernetes
35mEconomics of using Local Storage Attached to VMs on Cloud Providers [I]
19mPersistent Storage for Containers using Gluster
19mKeynote: Debunking the Myth: Kubernetes Storage is Hard
36mStorage Wars