DEV Community

Discussion on: Designing Video Architecture That Scales With Your Product (Not Against It)

Collapse
 
forasoft profile image
Fora Soft

Great breakdown on why SFUs are the go-to for real-time video.

From what we’ve seen at scale, just throwing an SFU at the problem only gets you so far. Once you move past small calls into big broadcast-style scenarios, things get tricky fast. In practice it usually looks like:

  • Up to ~50–100 people: plain SFU works fine
  • Hundreds: you start needing regional or cascaded SFUs
  • Thousands+ viewers: WebRTC ingest feeding transcoding and then CDN distribution

Simulcast or SVC is huge here too. Without adaptive layers, your SFU can hit a bandwidth wall fast.

Curious if the author has experience with cascading SFU architectures or edge SFU deployments?