I just started evaluating video conferencing frameworks including Jitsi for a possible project that resembles a remote classroom. It’s mainly a single person (instructor) video streaming to participants majority of the time. I am learning about video conferencing architectures MCU, SFU etc. (here iss something I found: https://webrtcglossary.com/sfu/) and I have a few questions.
Note that these may be premature but I am trying to find out if Jitsi can be used for this scenario or if I should consider something else alltogether. If I am not mistaken Jitsi seems to be a SFU, so here are the questions:
- MCU seems to be more suited to one person streams to N other persons case, however, looking at the MCU-SFU video explanation above, as long as the other parties are not sending any data it should not be much worse (traffic-wise)?
- Is it possible to close/stop upstream connections (or send no data at all) from other participants while the instructor is streaming to keep the traffic low? Is this a valid use case that I can implement, e.g. can I open / close participant upstream connections easily on demand? Or are they supposed to be open and streaming all the time. (This is with the assumption that N people streaming everything to each other all the time is just not feasible in my scenario where N could be large)
- Assuming the previous item is possible, would it give me enough performance for conferences of up to 100 people? Any judgment call on whether this would work by some more experienced users or developers of Jitsi?
- Any other ideas, stuff I am missing or misunderstood, or suggestions for frameworks more suited for my scenario?