Jitsi Architecture - pvgupta24/Jitsi-Meet-Concepts GitHub Wiki
Jitsi provides us a secure and easy to use video conferencing platform. There are a lot of services which work together to provide for the high quality real-time conferences.
Here'a a brief overview of how it works:
Jitsi Videobridge (JVB)
JVB is the heart of the conferences, which relays the media content to the connected participants instead of mixing them. It allows the conference to scale effectively.
In the context of a video conference, scalability means support for the growing number of people who join a conference. There are 3 major models which are widely used while deploying a multiparty video conference:
Peer-To-Peer (Mesh)
Each participant sends the media packets to all other participants connected on the webRTC data channel.
Because of being server-less, the mesh topology offers certain advantages:
- Low Cost
- Easy to implement
- Ensures data privacy (End-to-End encryption can be enforced for the transmission)
But as the number of participants increases, this leads to certain problems:
- Increased network traffic
- Does not Scale well beyond 3-5 participants
So, it becomes clear that we need a central server component as the number of participants in the conference increases where he participants and the server are connected in a star topology, with the server at its center. SFU and MCU are the popular alternative models used for better scalability than Mesh topology.