Load Balancing

How it works

A load balancer sits in front of a server pool and distributes incoming requests so no one machine is overwhelmed. Strategies range from round-robin to least-connections to weighted routing, with health checks that pull dead nodes out of rotation.

Mental models

• Round-robin, least-connections, IP-hash, and weighted are the staples.
• Health checks + automatic failover keep traffic off dead servers.
• L4 balances on IP/port (fast, just forwards packets); L7 reads HTTP paths, headers, and cookies to route smarter.
• SSL termination at the LB means backend servers skip the expensive decrypt/encrypt work.
• Horizontal scaling only works if servers are stateless — sessions belong in a shared store, not local memory.

Common pitfalls

• The load balancer itself is a new single point of failure — pair it with a standby in active-passive or active-active mode.
• Sticky sessions reintroduce state into a 'stateless' fleet; prefer a centralized session store (Redis, DB) instead.
• An underprovisioned or misconfigured LB becomes the bottleneck it was meant to prevent.