Self-Hosting LiveKit

Run the full stack on your infrastructure

Deploy and operate LiveKit server, agent workers, and supporting infrastructure on your own hardware. Covers when to self-host, Kubernetes deployment, agent worker scaling, monitoring, and security hardening.

What You Build

Production self-hosted LiveKit deployment with agent workers, monitoring, and security hardening.

Prerequisites

→Course 5.2

Chapters

Self-hosting architecture & decision framework

25m

When to self-host (data sovereignty, latency, cost at scale) vs when Cloud is simpler. Full architecture: LiveKit server, Redis coordination, TURN servers, agent workers, and monitoring. Hardware sizing for target concurrent sessions.

Cloud vs self-hostedArchitecture overviewHardware requirementsAgent worker deployment

Kubernetes deployment & server configuration

30m

Deploy with Helm charts on Kubernetes. Configure livekit.yaml (signaling, media ports, TLS, TURN). Set up Redis for multi-node coordination with persistence and Sentinel HA. Network requirements for WebRTC (UDP 50000-60000).

Helm chartsServer configNetwork requirementsTURN setup+1 more

Monitoring & security hardening

25m

Monitor with Prometheus metrics (room count, participant count, packet loss, CPU). Build Grafana dashboards and alerting rules. Harden security with Kubernetes network policies, TLS on all paths, API key rotation, and CORS configuration.

Prometheus metricsGrafana dashboardsAlerting rulesNetwork policies+2 more

Upgrades, backup & operations

20m

Rolling upgrades on Kubernetes with zero downtime. Back up Redis state, configs, and secrets. Recovery procedures for data loss, corruption, and full failure scenarios. Rollback with Helm.

Rolling upgradesBackup proceduresRecoveryMaintenance windows

What You Walk Away With

Ability to deploy and operate a self-hosted LiveKit installation with agent workers, Redis coordination, monitoring, security hardening, and upgrade procedures.