I'm building a webrtc project, and i need to:
- route specific stream to specific users
- record the streams on the server
I know this is typically the job of an SFU (selective forwarding unit) However, before finding out about SFUs, I had previously started using browsers running on servers (i tested both chrome and firefox...using firefox for now), and it seems to be working.
I run my javascript and create peer connections and add the relevant streams just like i would on the clients.
I was even able to successfuly achieve multi-server hierarchy this way.
The only downside is see right now, is that the browsers decode the streams, which i believe would cause cpu overhead which i would not see using a proper SFU.
However, my project generally does 1 to many streaming (or rather few to many), and i need server side recording (which would cause an SFU to decode the streams anyway)
So, my question is..
Why is using a browser as an SFU for webrtc a bad idea? I haven't seen a lot of people doing this, so there must be a reason
Thank you
