Blog Post

How Verizon Wants to Speed Up Your BitTorrent Videos

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!

p4p1.jpgVerizon and Pando will use today’s DCIA conference in New York to release test results of a system aimed at reducing the impact of P2P traffic on ISPs. AP reported about this first, promising that the system will make P2P downloads up to six times faster while at the same time reducing the costs for ISPs. CNet added a few more details about the technology known as “P4P,” including the obvious fact that you really won’t see much higher speeds with a basic DSL line that’s already maxed out by your downloads. FIOS customers, on the other hand, should be able to tell the difference.

P4P is the result of a joint working group that includes P2P companies and ISPs as well as researchers from Yale that have been working on optimizing P2P networks for years. Here’s how it works, and what it means for your torrent downloads:

The basic idea behind P4P is fairly simple: A dedicated server collects network topology data from ISPs and relays the information to P2P networks, optimizing specific transfers. The system is supposedly open to different P2P technology approaches, but it’s clearly designed with BitTorrent in mind.


P4P compared to traditional P2P. Source: DCIA.

BitTorrent uses a so-called tracker server to coordinate downloads of a specific file. A tracker will let you know who else is sharing the file you’re trying to download, and it will keep track of your upload data to make sure you’re contributing to the network. Trackers currently don’t care about the actual location of their users. P4P plans to change this by introducing a second tracker server to the system. This second server, dubbed iTracker, won’t keep track of the actual uploads and downloads, but of the network topology being used to facilitate them.

Let’s say you’re downloading a movie from Vuze’s tracker would first gather the IP addresses of the people sharing the movie and then query the iTracker server to see where these IP addresses are actually located. Are any of them using the same provider or the same backbone infrastructure? iTracker will help to find clients close to each other and get them to swap data within an ISP’s network.

The iTracker server would also know how a certain ISP wants to utilize its network at any given time. A certain link tends to be congested in the evening? iTracker will tell Vuze’s tracker to use a different set of connections for your download. Future iTrackers could be run by ISPs themselves, or by trusted third parties, with Pando’s CTO Laird Popkin suggesting a nonprofit like ARIN as a possible host.

So what will this mean for your Joost streams and Pirate Bay downloads? That depends, in part, on how ISPs will implement the system. Joost doesn’t rely on a tracker architecture, so it may be a while before it can actually make use of something like P4P. ISPs could also decide to set up P4P in a way that makes it impossible for trackers like the one run by The Pirate Bay to access the system, for example by running their own iTracker servers that only communicate with licensed P2P systems.

But there is a good chance that they won’t bother doing so. P4P can save them lots of money. Verizon, which made use of an iTracker run by Yale for its recent test of the technology, was able to route 56 percent of its P2P traffic locally with the help of P4P, whereas usually only 6 percent of all P2P transfers happen between Verizon customers. This means that they’ll have to buy much less bandwidth to give you access to your torrents.

Verizon or any other ISP implementing P4P would save even more money if they’d leave running the iTracker up to a third party that serves all BitTorrent trackers, such as the one run by, or the one run by The Pirate Bay. And ISPs like to save money, so the odds are good that your Pirate Bay torrents will get a boost from this as well.

12 Responses to “How Verizon Wants to Speed Up Your BitTorrent Videos”

  1. ThePositronicMan

    I’m afraid that there’s not much that can be done to speed up a dial-up connection… You may want to speak to your ISP about possibilities in this direction, such as compression for the data flow, but the best thing would be if you could get broadband somehow. You may be able to get it wirelessly, even if you can’t get cable or DSL, or a comparable wired technology.

    Also, if worst comes to worst, you may be abke to set up your own personal “bridge” to a high-speed connection, by a number of means, but this can be very difficult and/or time-consuming. Post any more questions here, and I’ll attempt to answer them as best I can, or direct you to people who would be able to better than I.

  2. “Several papers from academia have been published on this specific topic years ahead of the yale kids”

    PJ, there are several groups working on the general area of improving P2P. If you know of anyone that’s working on anything directly relevant to ISP’s and P2P companies that’s not already in the P4P Working Group, I’d love to hear about it. Please email me at [email protected].

  3. PJ CommonSense

    regardless of if verizon or bo jangles thinks it’s interesting it isn’t anything new nor was it called “p4p” when I was reading about it a few years ago.
    Also you think this is the first time people have done a “test” like this!!!??? obviously we’ve never heard of planetlab? go look on their consortium members list…
    But again who says doing a simple search on prior art is common sense cause we all know common sense is common right?

  4. Much of the technology used to deliver online video is has been around for a while. What’s important is that they are working with big business and ISPs to prove and implement them on a large scale.

  5. Pj CommonSense

    this is old tech. Several papers from academia have been published on this specific topic years ahead of the yale kids… it’s nice that their feeb education is catching up but this topic and these exact claims have already been proven in academia. Maybe their advisor should actually do a check for prior art before publishing findings that someone else already had. No der right?