Why architecture matters here

LB failures cause outages far larger than the backend problem. Architecture matters because health + algorithm + draining decide user impact.

Advertisement

The architecture: every piece explained

The top strip is the split. Client. L4 (TCP) flow hashing. L7 (HTTP) path aware. Backend pool.

The middle row is control. Health checks. Algorithms. Sticky sessions. Connection draining.

The lower rows are ops. TLS termination. Observability. Ops — failover + capacity + config.

Load balancer — L4 vs L7 + health checks + algorithms + draining + stickyroute traffic without dropping itClientrequestsL4 (TCP)flow hashingL7 (HTTP)path + header awareBackend poolhealthy targetsHealth checksactive + passiveAlgorithmsRR / least-conn / EWMASticky sessionscookie / IP hashConnection draininggraceful shutdownTLS terminationcert + SNIObservabilityper-backend metricsOps — LB failover + capacity + config safetyprobepickaffinitydrainterminatewatchwatchoperateoperate
Load balancer L4 vs L7 with health checks and draining.
Advertisement

End-to-end flow

End-to-end: client hits LB. Health checks maintain healthy set. Algorithm picks target. Sticky maintains user session. On deploy, connection draining lets in-flight complete. TLS terminated at LB.