P4P, IPV6 and RSS Player

[SawyerIII]

A couple of questions...

Are there plans to integrate whatever P4P protocol becomes the standard into the program?

If someone sent a piece file out on ipv6 to a IPV6 brodcast address, instead of a to a specific peer, wouldn't everyone who subscribed to that seeders IPV6 Broadcast address get that piece? If this is true, then wouldn't the list of IPV6 peers become able to send their pieces as Point-to-Multi-Point, and would this kind of interaction work with the current concepts of P4P.

Lastly, are there any expectations to integrate a full blown player and viewing tracker (as to whats been watched) into uTorrent (off the RSS lists)? I picture the ultimate podcast player, with an P4P IPV6 uTorrent controlling the transporting, viewing, storage, and deletion of video and audio RSS feeds like podcasts and TV feeds, we just need a Hulu-like desktop interface to the RSS feeds and files we are taking the time to download. uTorrent and RSS feeds should be able to make great offline replacement of Hulu, if it was done up.

-Robert

[DreadWingKnight]

Are there plans to integrate whatever P4P protocol becomes the standard into the program?

All the p4p research I've seen has been flawed, and there's a couple massive discussions already about it.

uTorrent will ONLY EVER SUPPORT TORRENT FILES. Regular podcast rss feeds won't ever be supported.

ISPs have this tendency of blocking multicast, hence why bittorrent exists in the first place.

[SawyerIII]

I understand, DreadWingKnight, that you never said you were nice, but I expected an intellegent response.

1. I have not heard of problems about P4P, do you have links about the discussion?

2. RSS is a subscription format, the files can be any file type from .mp3, & .avi to .torrent. With RSS as just the transport medium the possibility of selecting the .mp3 or .avi files from the .torrent AFTER it has been downloaded is a trivial difference. I.E. Once the files are downloaded, it is identical (with the difference of multiple files) to a podcast feed, if you add an element of selecting the cogent file.

3. Standard IPV4 multicast, yes, but what about IPV6 multicast (which does not send out to everybody, it just allows people to 'subscribe' to the address, which then has its packets routed to all the people on the list. Wouldn't this allow for the idea of 'Point-to-Multi-Point' torrents?

-Robert

[Ultima]

http://forum.utorrent.com/search.php?action=search&show_as=topics&keywords=p4p

[SawyerIII]

If P4P is so bad (which the suggested link(s) failed to demonstrate), then what about when it becomes part of the IETF?

The biggest complaint I read about P4P is that the local link may be slower that a distant link, therfore it is not a benifit to the client.

This does not make sense, as the demonstrations and the logic show that on a system with some high speed clients locally (FIOS, etc), can still make the system perform better for everyone.

Can you spell out your exact complaint with P4P, and be ready to discuss the real and potential pros and cons.

Recient P4P article...

http://gigaom.com/2009/11/11/p4p-may-be-coming-to-a-network-near-you/

[Switeck]

P4P results are based on both highly artificial tests (we will completely map out our ISP's infrastructure and specially program the client to take advantage of that versus a client that does not) and extremely optimistic assumptions (that torrents or equivalent will have LOCAL peers/seeds on the same ISP to connect to each other).

http://forum.utorrent.com/viewtopic.php?id=31519

[Ultima]

The biggest complaint I read about P4P is that the local link may be slower that a distant link, therfore it is not a benifit to the client.

This does not make sense, as the demonstrations and the logic show that on a system with some high speed clients locally (FIOS, etc), can still make the system perform better for everyone.

Explain what part of the "major" complaint doesn't make sense. Then explain to us how you can tell us that FiOS customers will get better results, and then magically extend those better results to all peers. How does FiOS customers getting better speeds benefit the swarm as a whole?

I don't profess to be an expert on the matter, but as I'm seeing it, P4P (if it works as claimed) creates artificial cliques in the swarm, potentially at the expense/detriment of the rest of the swarm.

[SawyerIII]

If you assume the swarm is fully discoverable for every client, then you would be correct.

The real world shows that p2p's biggest problem is related to accessing the entire swarm. If we can use a system (like iTracker) that will allow for swarms to INCREASE in size and focus on the fastest connections (which for FIOS users would probally be 2 hops away) then how could that be considered bad?

P4P is supposed to do 2 things: increase the visibility of the entire swarm & give a SUGGESTED metric for which connection should be used (odds are that they will start with the suggested ones first, and then head on to the others).

P4P can help discoverablility of clients on the network, which increases the available swarm size. Simple.

[DreadWingKnight]

The sweden effect pretty much counteracts any real benefit that p4p currently is supposed to give.

[SawyerIII]

The Sweden Effect?

[Switeck]

P4P can do nothing on any torrent with a small swarm.

P4P can do little on even a giant torrent with a medium-sized swarm (maybe 200 total peers/seeds).

P4P isn't *NEEDED* for the end user on the torrents with 1000+ peers/seeds.

It's only a gimmick so ISPs can use less transit bandwidth.

The Sweden Effect = extremely fast seeds/peers in Sweden means you won't be getting most of your data from the guy next door on the same ISP.