Chaos Engineering for SQL Server
summary
Demonstrates chaos experiments against a two-node Always On setup by stopping the database engine service and using PowerShell/Pester and T‑SQL to assert an availability group listener stays online, uncovering a default clustered role limit of one failure per six hours that blocked failback until changed. Shows SQL Server on Kubernetes with microk8s and an Azure cluster, uses KubeInvaders to kill pods, and confirms Kubernetes quickly recreates pods so the service recovers.
AI-written summary · about
chapters (11)
- 0:00About me
- 0:45Session Aim
- 1:47Agenda
- 2:39Identifying Weaknesses - Incident Analysis
- 4:04Likelihood - Impact Map
- 12:10Defining and Experiment
- 13:15Running an Experiment
- 14:45Demo
- 22:50SQL Server running on Kubernetes
- 24:14KubeInvaders
- 28:08Resources
related talks
44mChaos Engineering for SQL Server
32mImplementing Chaos Engineering in Kubernetes
57mSQL Server Containers & Kubernetes - Going to Production!
47mPods & Pages: Databases in Kubernetes- 57mSQL Server Containers & Kubernetes - Going to Production!
39mChaos: Breaking your Systems to Make them Unbreakable
25mChaos Engineering for Cloud native Apps
34mResilient Cloud App: Chaos Engineering, Azure Studio