Argo Rollouts at Scale: How We Brought Automated Rollback to 2100+ Micro-serv...
summary
Monzo replaces cluster-native rolling updates with Argo Rollouts across 2,100+ services, automating rollback as the default. A migration pipeline (manifest generator + shipper CLI + a migrator service) converts 2,143 deployments safely by batching migrations across the day, keeping deployment replicas predictable. After four months and ~7,000 Argo production deployments (22,000 staging) the system records 63 automatic rollbacks, reduces incident toil, and proves automated rollback at scale while preserving capacity and observability.
AI-written summary · about
chapters (14)
- 0:00Intro
- 1:36What is Monzo?
- 3:03Problem statement
- 5:36What is Argo Rollout?
- 7:54Why choosing Argo Rollout
- 9:06Is canary deployment the right thing for us? monzo
- 10:10Enter "Automated rollback"!
- 11:06Deployment flow
- 12:12Deployment architecture
- 14:23Refining our rules
- 15:44Take-aways
- 16:36How to migrate a service to Argo Rollout
- 18:04Automating the migration process away
- 23:50Operating argo-rollout at scale
related talks
29mTuning Argo Rollouts for Thousands of Workloads
24mArgo Rollouts from the Trenches – Next Level Progressive Delivery Best Practices
06m⚡ Lightning Talk: Improving Release Reliability with Argo Rollouts Across 10k Customer Services
30mProgressive Delivery Made Easy With Argo Rollouts
05mProject Update: Argo Rollouts Update
25mSailing Across the Ocean of Argo Ecosystem Like a Pro
27mUps, let's revert! 🔥 Automating Kubernetes Rollouts with Argo a...
46mFrom FTP to ArgoCD & Argo Rollouts: An adoption history inside a corp