Re:World’s Oldest Surviving Torrent Still Alive After 15 Years
by user5994461 on Sunday September 30, 2018 @08:04AM
(#18105817)
The Tor network has little capacity and it's abusively slow to use for normal web browsing. The experience for video streaming will be abysmal.
Re:World’s Oldest Surviving Torrent Still Alive After 15 Years
by crtasm on Sunday September 30, 2018 @10:46AM
(#18106369)
It's usually fast enough to watch YouTube, capacity has gone up a lot in the past few years IIRC.
Re:World’s Oldest Surviving Torrent Still Alive After 15 Years
by ThePhysicist on Sunday September 30, 2018 @09:09AM
(#18106026)
I'm not saying use the Tor network but rather implement the protocol on a custom, provider-specific infrastructure.
Re:World’s Oldest Surviving Torrent Still Alive After 15 Years
by Kalium on Sunday September 30, 2018 @11:39AM
(#18106607)
I believe your interlocutor is suggesting that the Tor protocol makes tradeoffs in its design that tend to make it slow. This is demonstrated by slowness in the largest Tor deployment to date.
Also, have a large and single-purpose Tor deployment controlled by a single provider would mean there would be no privacy from that provider and anyone able to send them subpoenas. I think that this weakens the privacy provided to the point where the decision to use Tor at all becomes questionable.
Re:World’s Oldest Surviving Torrent Still Alive After 15 Years
by ThePhysicist on Monday October 01, 2018 @04:34AM
(#18110901)
Sure but the goal of such a system is just to distribute content more efficiently so protecting the users from the provider of the system is not the goal, you just want to protect the users from each other, if you like.
Re:World’s Oldest Surviving Torrent Still Alive After 15 Years
by Kalium on Monday October 01, 2018 @09:28AM
(#18112321)
In that case, I'm afraid I don't see any advantages over a traditional CDN. Certainly not any privacy advantages, almost certainly not any performance advantages, and probably not any cost advantages either.
Is there something I've missed? Can you help me?
But you're probably right in that this is a relevant problem, as IP addresses are considered PII (in Europe at least), so leaking them could be problematic. That said, as Netflix already has a contract with each of its subscribers it shouldn't be too difficult to modify it to accommodate that kind of data sharing.
I imagine the real problem is ensuring streaming quality in such an infrastructure, as it's very hard to predict when a node will go down (e.g. user turns off laptop) or massively degrade in bandwidth (e.g. user starts downloading other content), and there's not much that customers hate more than interruptions when streaming movies (as you can probably attest firsthand). So given all these constraints it might be cheaper just using a CDN. Also, the bandwidth costs per customer probably make up a negligible share of overall costs (licensing is probably the biggest part) so it might not be worth optimizing this at all. I'd estimate that when I used Netflix I streamed maybe 20-30 movies / episodes per month (sometimes more), which would total at about 50-100 GB max (?) given the right encoding. For a company like Netflix that distributes probably petabytes of data a single GB should cost significantly less than 1 cent, so my bandwidth cost to Netflix would be between 0.5-1 $ per month (I even think it's much less probably).