Planning the Zero-Downtime Lifecycle of Your Service Mesh
summary
Shows how to upgrade an Envoy-based Istio mesh with zero downtime by running a canary control plane alongside the old one, rolling workloads and an ingress gateway over gradually, and rolling back on failure; recommends splitting ingress (or separate control planes) so ingress stays available, scoping config to namespaces and using Istio’s exportTo and envoyfilter version fields to limit blast radius, and warns CI/CD tools like Argo/Flux can disrupt upgrades.
AI-written summary · about
chapters (11)
- 0:00CHRISTIAN POSTA
- 2:08First ever Service Mesh Radar Report
- 3:28Service mesh is a critical piece of application networking infrastructure
- 4:39Loss of communication between control plane and data plane
- 6:31Loss of ingress to the mesh
- 7:22Bad configurations
- 8:19Complementary components for a service mesh
- 10:54Separate lifecycle in operator configurations
- 12:26Canary control plane and gateway
- 14:44Scope down configuration
- 16:52Scope down global configuration
related talks
38mIstio as an enabler for migrating Monolithic apps to Microservices
34mService Mesh for Developers
55mEmpower Your Microservices with Istio Service Mesh
26mIstio Migration: 50k+ Pods, Minimal Impact, Lessons Learned
38mFewer Moving Parts: Simplify Service Mesh Operations with Istio Ambient Mesh
37mZero-downtime Deployment Strategies (...) [ENG]
01h 06mIstio on Kubernetes: Smart Canaries, Dark Launches and Chaos
20mHow to Optimise your Mesh with Istio Sidecar