Pragmatic SRE for Kubernetes in the Cloud
summary
Pragmatic checklist of actions and costs: adopt SageMaker operators for Kubernetes to offload ML workloads, use EKS + App Mesh + X‑Ray + CloudWatch for tracing, and favor RDS for managed databases instead of running stateful DBs in-cluster. Instrument for fast detection and self-healing automation, centralize logs but budget for centralized log storage costs, train most of the team on the tools, and simulate common failures (CPU spikes, packet loss, region outages) rather than reflexively autoscaling.
AI-written summary · about
chapters (14)
- 0:00Pragmatic SRE for Kubernetes
- 1:14Introduction
- 2:18Detecting issues in Kubernetes
- 6:50Organized data for troubleshooting
- 11:05Tool selection and human factor
- 14:38Offloading workloads
- 16:27Target users and behavior
- 18:54Log storage costs
- 21:41System bottlenecks
- 23:28Common Kubernetes failures
- 25:55Availability vs cost
- 26:16Security risks in Kubernetes
- 29:04Emergency planning
- 30:08Q&A
related talks
56mBest practices for optimizing Kubernetes applications on AWS
22mHow to make Dev and Ops folks look good as an SRE
49mSite Reliability Engineering with Kubernetes
43mKubernetes cost optimisation - How to actually do it
33mKubernetes-Driven Multi-Cloud Deployments: Maximizing Scalability and Flexibility
33mKubernetes cost optimization - How to actually do it
14mCut Costs, Not Performance: Implement Kubernetes Cost Optimization Strategies
54mKubernetes for the konfused, a strategy toolkit for normals