News for hackers. Stuff that matters.
Re:The PeerTube content bootstrap fund by chrismorgan on Saturday May 16, 2020 @06:14AM (#23202182)
At the level of colocation and the likes, I believe it’s standard to just pay for the pipe and you can use it as heavily as you like. Some hosting providers extend that to their users, as well.

But there are practicality concerns to this for the specific domain of video distribution; suppose you’re a solo video producer: most of the time there’s a certain background level of watching, but when you release a new video there’s a massive spike of watchers. So in this case, having an enormously fat pipe and charging by data transferred is probably better for you than having a thin pipe that you can use as heavily as you like. If you have a 1Gbps link, for example, and 5Mbps video, you can never have even 200 people watching at any given time. But if you have a 40Gbps link and are charged by data transferred, you’ll be able to cope with a surge of closer to 8,000 viewers at once. (I bet that in practice it’d be struggling before even half that, but the point stands.)

I know some hosting colocation deals are done as a certain baseline with burst capacity, e.g. 1Gbps with 10Gbps burst capacity, which would mean something like “you can use up to 10Gbps, so long as 99% of the time you stay below 1Gbps”.

But all of these sorts of considerations are mostly important only when you’re a big shot, not when you’re just starting out or if you’re just doing it solo.

For myself, though, I’m more likely to accept B2’s ~1.5¢ per watch-hour rates, using it to host the actual video files (you can configure PeerTube thus), because they’ll have gigantic pipes. Then if I magically reach the point where that’s costing large sums of money, then I can optimise it.

There are no comments on this comment yet.