Design an L4/L7 load balancer that handles 5 million concurrent connections with sub-millisecond routing overhead and 3-second failover.
## Problem
Design a load balancer system capable of distributing traffic across thousands of backend servers at both the transport layer (L4) and application layer (L7). The system must handle millions of concurrent connections, detect and route around unhealthy backends within seconds, and support zero-downtime deployments through graceful connection draining. Think of building the infrastructure behind HAProxy, NGINX, or AWS ALB — but at hyperscale.
Sign up to access the full problem
Design canvas, rubric, hints, and model solutions.