Peer-To-peer Architecture Could Save Streaming Video
webrtcworld.com
Peer-To-peer Architecture Could Save Streaming Video
1–10 of 37 posts
Re: Peer-To-peer Architecture Could Save Streaming Video
#2We had peer to peer streaming 18 years ago. It does not scale for larger audiences, so if it stays on the 'skype' level then it will do fine, otherwise you're going to need some kind of redistribution network to handle the fan-out.
If you're willing to tolerate lag and decreasing fidelity you can also use peer-to-peer relaying but this only works for a very limited number of hops.
Re: Peer-To-peer Architecture Could Save Streaming Video
#3So... I'm conflicted about this. Sure, on one hand peer-to-peer streaming could solve some of the problems she's talking about, so long as there's a large enough audience for each video. Services like Steam have used P2P very effectively. At times it can break down to a very frustrating level quickly, so you'll still want CDNs around to act as seeds. It also doesn't work for live broadcasts which is becoming more common with the likes of Twitch. Also, the current technologies being discussed (marketed?) would require a plugin to work. I don't think this will fly until a browser-native solution is engineered.
Re: Peer-To-peer Architecture Could Save Streaming Video
#4A whole article on p2p video streaming, and yet Popcorn Time [1] is curiously absent. The flagship program that proves p2p movie watching is feasible, even comfortable, when delivered through BitTorrent doesn't get a mention. In a way the program already saved streaming video already.
We can only speculate whether the omission is conscious attempt at avoiding linking to risque programs, or does that indicate the article is copy-pasted press release of content providers.
[1] https://en.wikipedia.org/wiki/Popcorn_Time ; https://popcorntime.io/ seems to be the leading fork.
Re: Peer-To-peer Architecture Could Save Streaming Video
#5We had peer to peer streaming 18 years ago. It does not scale for larger audiences, so if it stays on the 'skype' level then it will do fine, otherwise you're going to need some kind of redistribution network to handle the fan-out. If you're willing to tolerate lag and decreasing fidelity you can also use peer-to-peer relaying but this only works for a very limited number of hops.
I'm also skeptical. Off the top of my head, if I had to make this work, I might try two things:
1 - While streaming normally from the CDN, also preload segments from peers. Complexity and client-bloat aside, you don't lose much trying (assuming the client has the bandwidth)
2 - Peering with folk that appear to be close. I feel that doing it by network address or ip->geo lookup might work well.
Re: Peer-To-peer Architecture Could Save Streaming Video
#6>Instead of one endpoint pulling down a single unique stream, the P2P approach allows simultaneous users to exchange video segments among themselves rather than each connecting to a server to do so.
Issues:
-- residential/mobile internet service has asymmetric upload/download speeds.
-- ISPs are suspicious of heavy uploads originating from residential homes and could be flagged as "servers" in violation of ISP's TOS or reclassified as "business tier pricing" thereby increasing the billing amount.
-- too much latency for live mega events (World Cup, etc) because of asymmetry mentioned above. (Your neighbor tweets that Spain scored the winning goal before your P2P stream received the last packets showing that it happened.) Latency for bittorrent is ok, but for live megaevents, no.
I don't think P2P down to residential is realistic. However, P2P between commercial entities looks more workable. If Akamai projects that their CDNs can't handle the next mega event, they contract with Amazon AWS CDNs to handle some of the spillover bandwidth. And/or work with ISPs (Comcast, FiOS edge servers) to act as CDNs. Yes, it involves competitors partnering with each other but this looks more feasible than coordinating residential internet service to deliver the extra bandwidth.
As an analogy, residential neighbors don't transfer supplemental electricity to each other but power stations in Canada might supply extra peak electricity to power stations in New York.
Re: Peer-To-peer Architecture Could Save Streaming Video
#7>Instead of one endpoint pulling down a single unique stream, the P2P approach allows simultaneous users to exchange video segments among themselves rather than each connecting to a server to do so. Issues: -- residential/mobile internet service has asymmetric upload/download speeds. -- ISPs are suspicious of heavy uploads originating from residential homes and could be flagged as "servers" in violation of ISP's TOS…
> I don't think P2P down to residential is realistic. However, P2P between commercial entities looks more workable.
The nice thing with P2P is that both model are not mutually exclusive. You can have a single peer that is effectively served by Akamai's CDN, and you can have another single peer that is served by you behind your poor ISP line; the protocol will work the same and automatically select the one that offers the better bandwith -- so people will in practice fetch from Akamai's CDN when it has capacity, and automatically switch to anybody else when Akamai alone isn't enough. You don't even need any logic, it will be handled automatically.
Re: Peer-To-peer Architecture Could Save Streaming Video
#8I'm not sure if that's going to work when more and more people are using mobile internet. Sure, we can all hope for real mobile flatrates but p2p videodistribution will add more traffic the networks … wait, no it wouldn't, would it?
Re: Peer-To-peer Architecture Could Save Streaming Video
#9So... I'm conflicted about this. Sure, on one hand peer-to-peer streaming could solve some of the problems she's talking about, so long as there's a large enough audience for each video. Services like Steam have used P2P very effectively. At times it can break down to a very frustrating level quickly, so you'll still want CDNs around to act as seeds. It also doesn't work for live broadcasts which is becoming more com…
> so long as there's a large enough audience for each video
Completely a non-issue. The providers would "seed" all the videos from their own servers, too, so when there aren't enough peers seeding it, the bulk of the streaming would be offered by the provider.
Using the Pareto principle, something like 80 percent of the videos would be "long-tail" with few to no seeds (other than from the provider), but only use 20 percent of the video traffic (so the provider's burden is greatly reduced anyway). The other 20 percent videos would represent 80 percent of the traffic, and those are the videos that would be helped most by P2P streaming.
Re: Peer-To-peer Architecture Could Save Streaming Video
#10>Instead of one endpoint pulling down a single unique stream, the P2P approach allows simultaneous users to exchange video segments among themselves rather than each connecting to a server to do so. Issues: -- residential/mobile internet service has asymmetric upload/download speeds. -- ISPs are suspicious of heavy uploads originating from residential homes and could be flagged as "servers" in violation of ISP's TOS…
You're thinking backwards of the problem. The ISPs don't offer high upload speeds because there's no demand for it. If there was, they would be forced to raise those speeds (in a competitive market, at least). Google Fiber offers symmetrical 1Gbps for download and upload speeds.