Service Mesh at Scale: How Xbox Cloud Gaming Secures 22k Pods with Linkerd
summary
Xbox Cloud Gaming runs Linkerd across 26 AKS clusters with ~50 microservices per cluster and 700–1,000 pods each, securing 22k pods with zero config mTLS and rolling cert rotation. Linkerd runs in HA mode, feeds Prometheus metrics used by Flagger for canary traffic splits and automated rollback, integrated into Prometheus + Flagger + Azure ADO pipelines. Outcome: thousands of dollars saved on monitoring, engineers freed from bespoke mTLS ops, and canary pitfalls identified for async and low‑volume workloads.
AI-written summary · about
chapters (14)
- 0:00Intro
- 1:50Footprint
- 3:45What is a Service Mesh
- 4:50Why did you want a Service Mesh
- 6:42Choosing a Service Mesh
- 8:35Why Linkerd
- 9:32How we are using Linkerd
- 13:00Observability
- 13:50Canary deployments
- 15:42Key learnings
- 17:59Engineering and cost savings
- 19:05Future plans
- 20:40Meet Abraham
- 21:15Outro
related talks
30mService Mesh Is Still Hard. The Things we all Did Wrong & Tales of Woe
34mService Mess to Service Mesh
40mWhen Failure is Not an Option: Processing Real Money at Monzo with Kubernetes and Linkerd
32mBeyond Security – Using Linkerd to Get Metrics
52mService Mess to Service Mesh
28mService Meshes: At What Cost?
22mCompliance the Easy Way: Zero-conf mTLS for Dev and smooth Day-2 f...
33mService Meshes: Istio, Linkerd - or No Mesh at All?