Deploying hybrid topologies with k8s & Envoy: a look at service discovery
summary
Migration moves service hosting from long‑boot VMs to Kubernetes with Envoy sidecars and a new Go control plane that streams service endpoints via gRPC. Scaling shifts from 15–20 minutes vs seconds for pod startup. Config merges legacy IP registries and k8s watches into unified Envoy configs so traffic can be shifted gradually. Observability gaps (missing IP tags) force pauses to add metrics. Built kill switches, health‑check failovers and an incremental rollout and rollback procedure that preserves production reliability.
AI-written summary · about
chapters (12)
- 0:00Start
- 2:31What is?
- 7:24Immutable Infrastructure
- 9:45What is envoy?
- 10:56Lyft's Topology: Universal Network Primitive
- 12:07The Mesh - what is it?
- 12:37What is Service Discovery?
- 13:25Pre-k8s Service Discovery
- 14:04Desired end state
- 16:40Hybrid Service Discovery
- 20:01Goals: achieved
- 20:36Lessons Learned
related talks
33mTake Envoy Beyond a K8s Service Mesh - to Legacy Bare Metal & VMs + More
25mBuilding Low Latency Topologies with Envoy
22mUnlock Kubernetes & Envoy: Master Your API Ecosystem!
25mApplication Networking on Kubernetes, where are we now?
15mKeynote: Deploying an Envoy-Based Service Mesh in Minutes Across Kubernetes and VMs
20mBuilding an Edge Control Plane with Kubernetes and Envoy
38mEnvoy Deep Dive
23mBridging the gap between on-prem and cloud: a story about...