Site Reliability Engineering with Kubernetes
summary
Practical SRE patterns for Kubernetes aiming at 99.99% uptime with concrete tooling and numbers rather than vague best practices. Uses Prometheus + Grafana for SLI/SLO dashboards (14 day sampling, 30 day maturity), Jaeger distributed tracing and Fluentd logging, Redis caching, separate etcd for performance, CRDs/operators and Go for automation, VPA/HPA plus Cluster Autoscaler for spikes, chaos tests in CI and multi‑region DR. Demonstrates a live Kafka streaming dashboard working end to end and recommends Argo CD for multi‑repo GitOps.
AI-written summary · about
chapters (9)
- 0:00Start
- 2:52Key areas of focus on K8s as an SRE
- 9:19Availability
- 16:05Reliability
- 19:26Observability
- 33:28GitOps CD - Automation
- 34:49Container Healthcheck
- 37:18Single point of failure
- 39:17Monitoring
related talks
22mHow to make Dev and Ops folks look good as an SRE
43mThe SRE's Kubernetes Production Checklist
24mSREs are Driving the Evolution of Developer Self-Service
09mObservability in Kubernetes: A Signal- and SLI-Centered Approach to Reliability
01h 02mDesigning an open-source observability stack for Kubernetes
32mStories from the Playbook
30mPragmatic SRE for Kubernetes in the Cloud
25mAn SLO-Driven Approach to Enhance Kubernetes Cluster Reliability