P2P Is Coming To YouTube

It looks like the idea of a P2P-powered YouTube is finally becoming reality, albeit without any contribution from Google. Singapore Shanghai-based P2P start-up PPLive, which we previously covered for its hugely successful P2P video platform, is experimenting with a P2P accelerator for Flash video streams. The application, which is dubbed PPVA, essentially distributes the stream of any popular Flash video from sites like YouTube via P2P without any involvement of the hosting server. [digg=http://digg.com/tech_news/P2P_is_Coming_to_YouTube]

PPLive began a Chinese-language only beta test of PPVA in June, and says it already clocked more than 10 million downloads, with the maximum number of simultaneous users being as high as 500,000. We tested PPVA with some popular YouTube videos, and the results are intriguing — especially if you consider what this could mean for online video hosters and content delivery networks alike.

PPVA, which is so far only available for Windows, is essentially a small plug-in that just sits in your task bar until it detects a Flash video stream. The app then finds out whether other PPVA users have accessed and cached the same clip; if that’s the case, it will request some of the data from them. A small status window shows where the data for each clip comes from, as well as other details.

So how well does it work? That really depends on the popularity of the clip in question. Access one of YouTube’s most popular videos, and only the first few bytes are requested from the server. After that, P2P distribution kicks in and that number quickly rises to 100 percent. Other popular videos show P2P distribution rates of around 30 percent, whereas more obscure stuff comes straight from YouTube’s servers.

PPVA also seems to work quite well with popular CollegeHumor videos, but there was no P2P distribution available on Vimeo or Blip. In fact, in some cases the plug-in seemed to make things worse, causing noticeable stuttering during playback of some clips, while crashing with others.

Those beta woes aside, the potential implications of something like PPVA are huge. Google could save a whole bunch of money on YouTube traffic without actually doing a thing, and smaller hosters could avoid embarrassing server downtime that so commonly occurs when a clip suddenly becomes popular. PPVA could also make P2P CDN offerings like BitTorrent’s DNA obsolete. After all, why would anyone pay for P2P content distribution if users can do it for free?

Of course, some content providers might be uneasy about not being asked whether they want their videos distributed via P2P. This becomes an even bigger issue when advertisers start requesting more detailed statistics about online video usage. PPLive told us that every video gets an initial request from the hosting server, which should allow video hosters to keep a tally of requests and viewers. But Google is reportedly moving towards a more detailed statistical analysis that looks at which parts of a video are being watched and which are skipped. PPVA could seriously distort these statistics.

Does that mean Google will get upset about PPLive’s experiment? Only if they’re hypocrites: Google invested $5 million in a Chinese P2P startup called Xunlei in late 2006. And guess what? Xunlei’s download manager is doing for files pretty much the same thing PPVA is doing for video streams, speeding up http downloads through P2P without consent of the original hosting provider.

PPLive has told me that an English-language version of the plug-in will be available in about a month.¬†Asked about ways to monetize PPVA, PPLive’s James Seng had this to say: “It’s new. It’s cool. We will figure out the rest later.”