Jitsi Videobridge

Jitsi Videobridge is the media routing engine at the core of Jitsi Meet. In a multi-party video call, every participant’s video and audio stream passes through the Videobridge, which forwards each stream to the other participants without decoding or re-encoding the media. This Selective Forwarding Unit (SFU) architecture keeps CPU usage low compared to older Multipoint Control Unit (MCU) designs that transcode every stream.

In practice, most people encounter Jitsi Videobridge as a component of a full Jitsi Meet deployment rather than as a standalone service. The bridge handles media; Jicofo (Jitsi Conference Focus) handles conference management; Prosody handles XMPP signalling. These three components together produce a working video conferencing platform. Running JVB in isolation makes sense only when building a custom WebRTC application that needs a self-hosted SFU backend.

Horizontal scaling is built into the architecture. Additional JVB instances can be added to a deployment and the load distributor routes conferences across them. This makes it possible to scale a self-hosted video platform to handle more simultaneous conferences without reconfiguring the signalling layer.

The Apache-2.0 licence means commercial use is unrestricted. Organisations running their own video infrastructure for compliance, data sovereignty, or cost reasons use Jitsi Meet and its Videobridge component to avoid reliance on commercial media services like Zoom, Teams, or Twilio Video.

The operational complexity is real. Java memory requirements are significant, UDP port ranges need to be open on the firewall, and TURN server configuration is required for participants behind restrictive NATs. For most self-hosters, the official Jitsi Meet Docker Compose setup handles this correctly and is the recommended starting point.