YouTube and Chrome created expectations of instantaneously loaded content, including HD videos. Unfortunately it seems the quality of a solution seems to be partly based on that, the time to the first playable frame. I don’t want to engage in this race but I admit it made me curious about the current architecture and how this is feasible already today.
I imagine enabling HLS and providing transcoding is the most impactful step. This way a video starts in e.g 480p then scales up in resolution. After that though I’m wondering if having closer to the audience redundancy could help. I started to wonder if
webtorrent-cli on servers seed boxes in different continents, closer to viewers, could help significantly. Then I started to wonder how to get the tracker … and found https://docs.joinpeertube.org/admin-following-instances?id=instances-redundancy and https://docs.joinpeertube.org/maintain-tools?id=peertube-redundancyjs
Did anybody replicated their PeerTube instance over several physically distant server (using e.g. How to migrate a peertube docker instance to another host ) and measured a significantly improved quality of experience for users?