Developing a Custom Load Balancer to serve traffic at scale
summary
Design and build a custom load balancer using ono as the dataplane and a Go control plane talking over gRPC, running as two Linux services on a single VM with configs stored in a DB and delivered dynamically (cloud‑init for provisioning). Demo shows round‑robin across two backend VMs, automatic health checks removing a dead node, and live config updates without restarts. Claims about cost show roughly ~50% lower hourly cost versus a small ALB at this scale.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:07What load balancing is
- 3:13Cloud-native options
- 4:56Envoy components
- 6:41Designing custom LB
- 8:48Dynamic config setup
- 11:13Cloud-init deployment
- 12:17Live demo
- 13:41Cost savings
- 15:28Q&A
related talks
33mCloud Native Load Balancing From Scratch
33mCloud Native Load Balancing From Scratch
40mGood, Bad, and Ugly - the Art of Load Balancing at Scale
24mLoad-Balancing Node.js in Production
22mLoad Balancing in gRPC
37mBuilding the Multi-Cluster L7 Load Balancer
25mLoad Balancing in gRPC
23mBuilding Unified Global Load Balancing for the Edge With k8gb