They talk about stress-testing of livestreaming towards the end of the article. A quick search shows that they've done several of them for videos as well. The peer-to-peer structure really protects them from sudden scale in demand.
In theory, for sure. But in reality the p2p part might not work for lots of users for various reasons (firewall etc.) so then the server gets all the load and chokes. At least that has be my experience when experimenting with webRTC.
"Last year, thanks to French indie journalist David Dufresne’s Au Poste ! livestream show and his hoster Octopuce, we got a livestream stress test with more than 400 simultaneous viewers "
And 400 is not a very big number, if one has ambitious goals. But it is a good start.
So, what if peertube was a good answer to 99% cases ?
I ask because that's the kind of runway where there's time to continue to work out questions like this, and they aren't at present make or break, and so far as I can tell there's no reason to believe issues about firewalls pose an issue that's any more or less difficult than the technical challenges they've already solved.
Certainly. I am 100% certain that the whole concept is doable.