Pods & Pages: Databases in Kubernetes
summary
Running SQL Server on Kubernetes claims fast recoverable deployments and high availability but warns against monster multi‑TB IOPS workloads; a demo cluster on a three-node Azure cluster uses StatefulSets, PVCs, an Azure CSI storageclass, liveliness probes, tolerations and a Guaranteed QoS class, and shows 2 minutes to spin up a full SQL instance versus 45–60 minutes for VM installs, saving roughly a month of dev time per year. Volume snapshots restore a crash‑consistent data file but volume snapshots (crash vs application-consistent) matter — app‑consistent snapshot demo fails and operators are recommended for automated AG failover.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 2:56Why run databases in Kubernetes
- 9:23DBA responsibilities
- 15:46High availability for databases
- 22:13Database recoverability and backups
- 31:27SQL Server in Kubernetes demo
- 36:31Resiliency testing
- 41:47Q&A
related talks
57mSQL Server Containers & Kubernetes - Going to Production!
42mImproving Availability for Stateful Applications in Kubernetes
45mDatabases on Kubernetes: Why you should care- 57mSQL Server Containers & Kubernetes - Going to Production!
35mStateful Apps On Kubernetes - Bring Them On!
30mChaos Engineering for SQL Server
26mChallenges of Managing Stateful Applications in Kubernetes
06m⚡ Lightning Talk: Disaster Recovery - How IaCaC and Kubernetes Enables Cost Efficiency and Fast Recovery