Ready to Go - Measuring Release Fitness at Scale
summary
Adidas runs 10,602 services across 1,345 namespaces in 37 Kubernetes clusters and builds a single go/no-go signal that automatically blocks production promotions when KPIs fail. The system aggregates SLOs and error budget, CI/CD health from Jenkins + Prometheus + Grafana and availability checks (blackbox exporter, Elastic/HOMES), plus value‑stream dependencies and regression status. The dashboard computes a binary outcome and the promotion pipeline enforces it, reducing manual release gating.
AI-written summary · about
chapters (6)
- 0:00Introduction and growth
- 4:59Release fitness concept
- 8:10Building the fitness function
- 15:17Automating the dashboard
- 19:53Using the dashboard in CI/CD
- 21:41Scaling and next steps
related talks
34mThe Journey of Adidas to a Global Kubernetes Rollout
33mScaling k8s Deployments: Achieving Elite DORA Metrics
13mDevOps automation with Go
31mRelease Engineering as a Service - The Art of Integrating with K8s
31mRelease Engineering as a Service - The Art of Integrating with K8s
34mKubernetes from Dev to Prod at GoEuro
21mQuality Gates in Production: How We Turn OpenTelemetry...
28mChaos Validation Made Easy