News P4P: The future of downloading?

http://www.bit-tech.net/news/2008/03/17/p4p_the_future_of_downloading/1

American ISP Verizon has become a charter member of the P4P Workgroup, a collective of ISPs with an interesting idea on how to make P2P perform better and cost less.

 

I'm all for transfer speeds increasing - my worry would be that if this is implemented does that then effectively give ISPs an excuse to keep records of P2P traffic?

 

To me it's just a great example of synergy. As long as they don't use it to pull surveillance into the deal but strictly try to optimize bittorrent efficiency then it's all good news.

 

as long as the media is cheap enough and you are not flooded by ads then i am all in.....

 

if u cant beat em, join em.

 

they already do that.

 

Yes, however, with the up-coming popularity of encryption between peers, I'm sure they'd love to have a nice work-around to know who's requesting which pieces (of what) from whom.

 

Color me cynical.

This would be great news is the ISPs wern't all money grubbing whores who were owned by content producers and therefore have an interest in controlling what content you can access.

It wouldn't be a problem if ISPs were just ISPs. unfortunatly all the broadband ISPs also want to be TV and phone providers and so they have a vested interest in not allowing you to download competing content that might keep you from buying their other services. That, plus the fact that they have shown a distinct willingness to roll over for the government on providing customer information, means that i want them as involved as LITTLE as possible in my web expierience.

 

This concept is straight out of the CCNA curriculum and the idea (or similar function) was already conceived by Cisco Systems Inc. ten or more years ago and is already partially implemented by corporate networks today. The problem with this system is that the 'broadcast' of network health doesn't really save much network congestion at all and can cause congestion on very large networks, like the Internet. This is because the Internet is configured in such a way that the broadcast might echo back and forth between the same recipients (or within the same subnetworks) before it reached its hop limit and expire--basically, not all routing hubs are 'informed' as needed. The additional resources required to 'map out' the massive network would have been better spent on simply expanding or improving existing routing paths.

 

Amon, P4P takes a very different approach from a router level 'broadcast of network health'. What P4P does is provide a standard way for ISP's to tell P2P systems how to tell which IP address are near each other in the network, so that the P2P networks can preferentially connect those peers. As such, it's a very low overhead application level protocol.

The problem that P4P addresses is the inefficiency of random peer assignments as used by most p2p systems. For example, imagine that you're downloading a file from a swarm with 10,000 peers, one of whom is right next to you. Using standard BitTorrent you would be 99.5% likely not to find out about the best peer on the first announce, and only 50% likely to find out about the best peer in 200 minutes, and have no guarantee of ever finding out about the best peer. With P4P, the network knows that you two are next to each other, so you're connected immediately. The result is that with P4P you download most data from peers that you're well connected with, resulting in 200% speedups (on average) for FTTH users compared to standard P2P. It also results in a 50% reduction in inter-ISP data transfers, which ISP's like. So we have a happy situation where this improvement in efficiency improves performance for users, and reduces costs to ISP's.

- Laird Popkin, CTO, Pando Networks and Co-Chair of P4P Working Group

 

Welcome aboard, Laird. It sounds like a neat thechnology, but the industry is going to have to answer privacy and monitoring concerns before I see this really taking off.

 

That's an important point, Cthippo - P4P has an intermediary between the ISP and the P2P network in order to protect the privacy of both. So the P2P network doesn't get a copy of the ISP's network map, just 'hints' to help figure out what IP addresses are near each other. And the ISP doesn't get any information about what's going on in the P2P network.

 

Ah, that clears it up a bit for me, then. Are there any necessary changes to or replacement of existing routing devices, or modification of the network structure? I'm not unconvinced of the advantage of P4P, but I'm a bit skeptical that such a drastic improvement can be produced on the Net's existing infrastructure.

 

P4P is entirely application level - there are no changes to the networking infrastructure. So implementing it is a lot easier thana network level approach.