There is a comical side to the P2P controversy. The ISP philanthropists are whining, music and movie publishers are suing, users cry foul, The Electronic Frontier Foundation (EFF ) jumps to the defense of the oppressed and raises money, pundits gravely contemplate the future of the Net and, of course, smelling a photo (or Web) opportunity, politicians strike poses.
What’s the story here, do we have a dog in that fight?
A quick reminder: P2P stands for Peer To Peer communication, as opposed to the classical client (my PC) server (amazon.com, hotmail.com) configuration. With P2P, users, peers, share files without central server. Using software such as LimeWire on your PC, you share what you have and get what you want. As usual, Wikipedia provides a robust explanation of the ins and outs of P2P. Let’s just say the magic of IP packets makes it work well, much too well.
The first to complain are the owners/publishers of content: for them, P2P is a way to propagate stolen property, it stands for Pirate-To-Pirate. The highs and lows of fights between the RIAA or the MPAA are well documented: google (it’s now a verb) “RIAA lawsuit” or “MPAA lawsuit” for more. How the “majors” think they prevail in the end remains a mystery. They should read Jonathan Swift, it’s free, in the public domain.
Next, we have the ISP. Yes, the potential for P2P was right there in the Internet basic technology, but nobody saw it coming. There are times where P2P consumes 90% of all Internet traffic. Design a finely calibrated business model for your All You Can buffet and, suddenly, a bus full of sumo wrestlers pulls in your parking lot. Comcast and others tried to choke P2P traffic. The result? More of an already bad reputation for gaming their customers and more politicians rooting for an exploit. As recent tests have established, the very sophistication of IP protocols that makes P2P possible also makes it very hard to distinguish a “bad” P2P transmission from a “good” exchange with schibsted.com.
But there is another side to P2P, one that saves money. Consider CDN, Content Delivery Networks. One example is Akamai, $630M in sales, 15% profit margin. One customer is Apple, the company “pushes” more than 100 megabytes of software updates per month, per user as well as large servings of content such as video tutorials or Steve’s much consumed keynotes. All free. Instead of owning and operating servers for that task, Apple offloads it to Akamai. That company, in turn, maintains a network of strategically located and finely tuned servers, hence the CDN moniker. Bandwidth is expensive, before being saved by Google late 2006, YouTube spent about $1M a month feeding free videos.
This is where the “good” P2P comes in. Invented in computer science labs at Stanford, variations of P2P automatically distribute content to users who, in turn, propagate it to others on request. “All it needs”, an admittedly dangerous phrase, is cooperating software on PC and a directory that keeps track of the propagating content. The CDN is hierarchical. By contrast, this modified P2P is a mesh network, one where bandwidth is shared between users. One important addition: with P2P, fragments of content come from multiple simultaneous connections, software re-assembles the load on my machine.
This matters to publishers and advertisers. A video on my site gains sudden popularity, Hillary says something true, my servers get “hammered”, I lose advertising revenue, or I have to pay a hefty CDN bill. If I want to deliver movie trailers, a form of advertising, or slick video eye candy to sell cars in Spring, or Summer fashion, or Greek islands vacations, the bandwidth expense hurts my business.
That’s the dog we, users, content designers and publishers, advertisers have in that fight. With a browser plug-in, my computer becomes part of a mesh CDN. Pay for my participation with goodies, discounts, freebies and we’re in business. As for the ISP, it won’t change the amount of traffic they carry and I trust them to invent some freemium variant.
The rest is, as engineers like to joke, a mere matter of implementation.