Home / Resources / Mesh vs SFU video calls
BLOG
Mesh vs SFU: why group video calls break at scale
Two ways to run a video call and when each makes sense.
The short answer
In a mesh call every person sends their video to every other person. In an SFU call each person sends one copy to a media server, which forwards it. Mesh needs no server but stops scaling quickly; an SFU costs a server but handles bigger rooms.
Side by side
| Mesh | SFU | |
|---|---|---|
| Upload per person | Grows with each added person | One copy |
| Server needed | No | Yes |
| Best for | Small calls | Bigger rooms |
| Cost | None | Server or a hosted allowance |
What Dock does
Small calls run peer to peer. For bigger meetings you can connect a LiveKit server. The app shows participant-minutes used against the free allowance you set.
Estimate your own
The bandwidth estimator shows upload per person for both models.
Questions
What is LiveKit?
An open-source real-time video platform that can be self-hosted or used as a hosted service.
Why not always use an SFU?
It needs a server and, on hosted plans, a bill.