Load balancing is the technique of spreading incoming network requests across multiple backend instances of the same service. The load balancer can be a dedicated appliance, a cloud service, a reverse proxy (nginx, HAProxy, Caddy, Traefik), or a DNS-level round-robin. Common algorithms include round-robin, least-connections, IP-hash, and consistent-hash. Load balancing is the operational mechanism that turns redundancy into both higher capacity (more concurrent users than one box can serve) and higher availability (one backend can fail without dropping every in-flight request).

In a self-hosting context

Self-hosters rarely need an explicit load balancer for a single VPS deployment of Mattermost, Nextcloud, or Plausible Analytics. A reverse proxy in front of a single backend is enough. Load balancing enters the picture when you scale to multiple backend hosts — at which point the same reverse-proxy server you already use (Reverse proxy) usually grows into the load-balancing role, with no separate component needed.