11 Comments

Summary:

Researchers at The University of Washington and Yale University will present a paper today on a developing Internet protocol that could lessen bandwidth demands from video and other large files. The peer-4-peer protocol is being touted by Pando Networks and a handful of ISPs as a […]

Researchers at The University of Washington and Yale University will present a paper today on a developing Internet protocol that could lessen bandwidth demands from video and other large files. The peer-4-peer protocol is being touted by Pando Networks and a handful of ISPs as a way to solve some of the traffic problems caused by peer-to-peer file-sharing services such as BitTorrent.

Compressing or managing data more efficiently is becoming increasingly important, as providers attempt to clamp down on large amounts of traffic and as consumers and corporations demand ever more bandwidth-intensive applications.

Like P2P, the P4P protocol breaks files up into smaller packets, sends those around the Internet and then reassembles them at a destination, but P4P tracks the most efficient point in the networks from which to swap those files. This involves the ISPs handing over information about their network topology and knowing where a file swapper sits on the network. P4P makes it possible to know these things without exposing the data to either side. In tests with Verizon, Pando showed that by using P4P it could increase delivery speeds by up to 235 percent on U.S. cable networks and reduce intra-network traffic by 34 percent.

On their own, such protocols can help, but they won’t stave off the need to build out more network capacity or make existing protocols more efficient. However, as carriers cry uncle under the supposedly heavy loads that file-sharing and P2P services put on their networks, seeking tiered services, bandwidth caps and other practices to handle the bandwidth hogs, an industry-blessed protocol may be welcomed.

image courtesy of Pando Networks

You’re subscribed! If you like, you can update your settings

  1. P4P only works with a “bittorrent compatible” protocol. Basically you put a “iTracker” in a ISP and it negotiates peers based on ISP network topology. It’s a broken design, unless you think that Bittorrent is the only P2P protocol both now and in the future for handling large volume data sets.

  2. Whether it’s P2P or the too-clever-by-half P4P, consumers are going to begin figuring out that they are being asked to pay storage and upstream bandwidth costs so that content distributors don’t have to. While college kids in dorm rooms might be ok with that bargain, mass market consumers will be troubled that their disk drives are spinning and upstream bandwidth is being used so that someone else can watch a movie.

    As many CDNs have shown, distributing storage is a good thing, but not necessarily all the way to the endpoint. A better approach, both in terms of storage and bandwidth costs (and, for that matter, energy costs) is to distribute content to the edge of the CDN and then provide QOS-specific channels over the broadband network. Consumers pay for QOS channels (if they want them), receiving DVD-like playback of HD content, and content only traverses the broadband network once, as opposed to twice for P-fill-in-the-blank-P distribution models.

  3. Luca Mondini » P4P: Provider Portal for Applications Thursday, August 21, 2008

    [...] potrebbero agevolare il traffico dati, che è ormai dominato proprio da applicazioni P2P. GigaOM riporta oggi in un post, il protocollo P4P: Like P2P, the P4P protocol breaks files up into smaller packets, [...]

  4. There are a couple misleading statements here, both by the author and in the comments that show a misunderstanding of P4P.

    1) P4P is not a P2P protocol. It does not say anything about how data is “broken into smaller packets”, how to manage peers, or what protocols to use to make connections.

    2) P4P is not just for torrent-based systems. Although Pando and many other systems out there are BitTorrent derivatives, there’s no reason a fully proprietary system cannot take advantage of P4P.

    P4P is a mechanism that allows ISPs to provide guidance on how peers in their networks can be efficiently connected, without sharing exact details of network topology. P2P systems can choose to ignore this data, but when used to guide P2P connections, it can improve localization and reduce painful off-net bandwidth. This ISP ‘guidance’ can take the form of a table that expresses connection preferences between various IP address ranges.

    Of course there are other ways to improve traffic localization without P4P support as well, and we are just starting to see these approaches from commercial P2P systems.

  5. I read the research paper and checked the experimental results. I feel this is just a project for a few more publications, not very useful, but not completely useless either. I don’t believe this will bring any significant change to CDN.

    “P4P is a mechanism that allows ISPs to provide guidance on how peers in
    their networks can be efficiently connected”

    Really smart P2P software can figure what’s the best route out on the fly. ISP’s guidance works for ISP’s interests, not always peers. Verizon sure has a lot of money to spend around no such projects : LOL

  6. @ random_graph
    It was tested on planet lab with the bittorrent protocol and only the bittorrent protocol. The same goes for the fields tests. This working with any other p2p technology is pure paper theory.

  7. P4P… less throttling of Broadband?! » Hiconomics Tuesday, August 26, 2008

    [...] into resolving the issues of P2P throttling ISP broadband capacity… GigaOM summarises it well here… but raises the fact that it will only resolve issues temporarily. But isn’t it all [...]

  8. Hulu is Bad for the Net but People Will Pay for Twitter – GigaOM Wednesday, September 3, 2008

    [...] content is such an inefficient way of getting video from one place to another that sites should find better options such as faster than real time streaming and buffering. The goal of the paper is to counter carrier [...]

  9. UW CSE News » Widespread coverage of P4P (August 2008) Tuesday, September 16, 2008

    [...] “Can P4P Solve Bandwidth Bloat? (GIGaom) (8/21/08) [...]

  10. Will P2P Soon Be the Scourge of Mobile Networks? Tuesday, July 21, 2009

    [...] July 21, 2009 | 5:30 AM PT | 0 comments Wired Internet service providers, by and large, view peer-to-peer file-sharing as a voracious, bandwidth-eating monster. With that in mind, they have justified tiered pricing, aggressive traffic shaping and [...]

Comments have been disabled for this post