Software Rollout at Scale: Using GitOps to Scale Kubernetes Rollouts
summary
Shows an opinionated GitOps pattern that scales from one cluster to ~100 managed clusters using Flux and a demo on kind, deploying podinfo plus an NGINX ingress. Proposes splitting infrastructure into its own repo, using git tags for per-cluster pinned versions and branch promotion path (innovators → early adopters → majority) with automated PRs to promote releases. Demonstrates how a single bad commit broke all clusters, then how branches and automation limit blast radius and provide visibility for customers.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:44GitOps fundamentals
- 4:38GitOps vs traditional CI
- 6:56Managing infrastructure
- 9:35Simple GitOps demonstration
- 15:31Multi-cluster scaling
- 22:24Versioned rollouts
- 25:38Adoption theory approach
- 30:18Branch-based promotion demo
- 36:40Conclusion and best practices
related talks
45mFrom DevOps to GitOps: Supercharging Your Kubernetes Workload Deployments
05mApplication promotion in GitOps
26mGitOps Lifecycle Management
35mPanel: GitOps User Stories
26mKeynote Session: GitOps as an Evolution of Kubernetes
26mEmpowering SRE with GitOps
27mLightning Talk: GitOps, A Slightly Realistic Situation on Kubernetes with Flux
23mGitOps: yea or nay?