| That’s why we have been building https://qbix.com/platform To have an open source alternative. Want videoconferencing on your own site? You can! See here for instance. https://yang2020.app/meeting We have a harder challenge of making all the SDP offers work cross browser, but Chrome should def work. Code: https://github.com/Qbix
(If you like it, star it lol ⭐️) Contact me if you want to learn how to use the Qbix platform. I will be teaching classes and put it online. We are following the wordpress model. My email is in https://qbix.com/about Quick question for the networking experts here... with everyone connecting from home, what percentage are behind a LAN firewall that you need to use TURN servers? What if you avoided those servers and made peer to peer infra entirely, how many people would we lose? (Is a complete graph of everyone sending to everyone worse than an SFU once you get too many users? Isn’t it exactly the same number of streams, just in a star topology? Can’t we just nominate a few of the browsers to do what the SFU does, namely forwarding video to the others? Is the issue only with resolution?) |
From home? Essentially 100%.
that you need to use TURN servers?
That's less clear. I'm not sure how many home firewalls are impenetrable by STUN as well. I worked on Twilio's WebRTC-based audio product back in 2012-2014. In the beginning we only supported STUN. We did get some customer support requests about initial connection failures (which I mostly attributed to STUN failures), but never kept track of stats on what the success/fail ratio was. We eventually added TURN support (after I left that product team), but based on how long it took us to do that, my guess would be STUN was effective for most setups. Also consider that many (most?) of our users were probably behind restrictive corporate firewalls, and I'd expect home firewalls to be more lenient.