Riddle me this -- what do you get when you cross the cost savings of a peer-to-peer network with the stability, reliability, and security of a traditional CDN (Content Delivery Network)? The answer, if one Seattle startup has its way, might just be on-demand DVD-quality video at a fraction of the current cost.
Sitting atop Seattle's famed FX McRory's building in the picturesque Pioneer Square, a group of 15 dedicated employees at GridNetworks is attempting to go where many have ventured but few have succeeded; they're attempting to morph the concept of peer-to-peer into a business.
Last week I sat down with GridNetworks CEO, Jeff Payne, and VP of Sales and Marketing, Bo Wandell, to get the lowdown on their new service. Click on to find out what I gleaned from their vision for Grid.
So what is GridNetworks?
The short version is that GridNetworks is one of a new breed of hybrid content delivery platforms built upon the principles of peer-to-peer. The concept is simple. Build out a traditional CDN, whereby content is propagated out to a collection of strategically placed edge servers, and then take it a step further. Instead of caching the content strictly on those servers, use the client nodes to house and serve the content as well. Whenever possible, have clients pull content from the grid of other clients. The traditional servers are then used to fill in gaps and push content onto the grid. Such a strategy helps to eliminate choke points and greatly reduces the bandwidth overhead required by a traditional system.
Do we need such an approach? Consider for a moment what the full transition to internet-delivered entertainment would actually entail. To do this we'll use some "back of the envelope" math. And, yes, these numbers are extreme cases, but, really, what fun is envelope math without extreme cases?
First, we'll start with movies. As a point of reference Netflix ships about one million movies a day to its customers. That's approximately 2 million hours of content per day. What would it look like if Netflix stopped shipping discs and started delivering purely over the Internet? For now we'll disregard HD movies and we'll assume a reasonable bitrate of 1.5Mbps. This equates to 1.5 Petabytes of data a day just to switch Netflix to internet delivery. This may seem like a lot, but really it's nothing compared to the big daddy that is the concept of completely internet-driven television.
According to the latest Nielsen reports, each of the over 100 million US households watches an average of 8 hours of television per day. That's 800 million hours of content viewed per day. Alternatively stated, that's about 500 Petabytes of data per day that would need to be delivered. What's worse is that these numbers are highly skewed towards peak times. During an average primetime lineup, well over 50 million households are tuned to just a handful of programs. To stream Desperate Housewives, for instance, would require that a traditional CDN push out nearly 4 Petabytes per hour for 3 hours. How much data is that? Well... it's enough to completely saturate nearly 1000 OC-192 lines (or 6.4 million T1s). In other words, it's a lot. And heck, that's just the US market.
In contrast, the folks at GridNetworks, working in conjunction with content providers, are confident that they could get the CDN portion of the above scenario to well under 400 Terabytes. And, even more importantly, by pushing the data to the end-user grid in advance they would be able to lower the peak bandwidth requirements by at least another order of magnitude. The result is an extreme cost savings for the content providers.
Why does that matter to you? That's simple. It means more content. You see -- there are vast libraries of content sitting in vaults blocked by (among other things) delivery costs. It stands to reason that if an episode of "It's Like, You Know" costs 50 cents to deliver, it's got to sell for more than 50 cents, and, as great as the show was, few people are going to pay more than 50 cents to watch an episode. If, however, you can cut those delivery costs down to 5 or 10 cents, there's a whole new market to be had.
That's not to say it's a clear road for GridNetworks. There are major obstacles still to overcome. First, and foremost, they've got to conquer the "new last mile;" they've got to get their technology to the television. While they've done their technical work by building atop a microkernel, convincing set top box manufacturers to incorporate their platform will be an uphill battle. Likewise, getting the major cable companies to ship with such a feature enabled will also be difficult.
Another potential stumbling block is convincing content holders that their solution is "safe." To that end, GridNetworks encrypts each segment of content at 448bit (Blowfish algo). What's more -- decrypting a segment (a feat that's yet to be done) would only get you a 6 second clip. Payne tells me that at least one studio has vetted and passed the security measure. He's confident more will follow suit.
Is GridNetworks the answer? It's unclear; we'll only truly know how well it will all work when critical mass is achieved. Until then, it's an interesting concept.
If you'd like to check it out, they've got a host of trailers on their site. It requires a download and, as such, you become part of the grid. However, a week of testing provided nary a blip in my VoIP service. So, they're either good bandwidth citizens or I'm the only one on the grid. Either way -- it worked fine.
If you have comments or suggestions for future columns, drop me a line at firstname.lastname@example.org.