February 9, 2011 12:26 PM
Netflix Proves America's Broadband Inadequacies
Netflix, purveyor of DVDs by mail and increasingly streaming video over the Internet, has released data highlighting the broadband capabilities of their users, and the results are grim.
While generally speaking US broadband providers are delivering something resembling their promised peak speeds, where they're coming up short are their sustained speeds.
The peak vs. sustained issue gets back to the fact that almost all providers sell broadband at an "up to" speed. So when you pay for 5Mbps service, they're not saying you're going to get 5Mbps all the time. Instead they're promising that sometimes your service will get as high as 5Mbps.
With many Internet applications, having broadband speeds vary isn't a big deal, like downloading a large attachment to an email. But with Internet applications like streaming video it's the sustained speed that arguably matters more than the peak.
This being said, the delivery of Internet video has evolved to overcome the limitations of "up to" broadband services through capabilities like variable bitrate streaming. What these kinds of technologies can do is adjust the bitrate (essentially the resolution or picture quality) of the video based on how much bandwidth is available.
So you may start out watching a video in HD at 5Mbps, but then when your available bandwidth drops the picture quality will worsen so that your stream doesn't stop and start getting all herky jerky.
What this means is that if you want to be able to watch a movie in HD on the Internet through Netflix, you don't just need broadband with a peak of 5Mbps (their HD video's encoded at 4.8Mbps), you need broadband that can sustain a 5Mbps connection. And this is where the results of Netflix's analysis turn grim.
What they've shown is that NONE of the major US broadband providers are delivering a sustained 5Mbps connection.
Now, I do have to admit that this isn't necessarily a surprise. The business model of residential broadband is based on being able to oversubscribe a backhaul connection to the Internet to make the cost of bandwidth more affordable to end users. Plus there are a number of other factors that could lower the sustained bandwidth being realized by a Netflix user, even basic things like someone else being in the same house on the same connection doing something else at the same time.
But I think it's a mistake to downplay the significance of these findings. The reason why is that there's a whole new class of applications that will require high rates of sustained bandwidth in order to work, the best example of which is videoconferencing.
There are no work arounds to delivering a high quality videoconferencing experience. With movies you can just have people download instead of stream it, so the sustained throughput is less relevant. But with a videoconference you need for both ends of that signal to be able to have adequate sustained bandwidth to support high quality video being send to and from each user.
Currently HD videoconferencing really only works on business-class broadband that has quality of service guarantees so that it can have enough sustained bandwidth to deliver a rock solid two-way video stream. But if we want to move those capabilities into the home, we need a broadband infrastructure capable of delivering sustained throughput.
One of my biggest fears is that our policymakers in DC and in state capitols across this nation don't understand this simple fact. They think that if they can get 5Mbps to people, that's good enough. But what they fail to recognize is that just getting service that delivers a peak of 5Mbps isn't good enough. What we need is broadband that's capable of sustaining those kinds of speeds.
What this Netflix data shows is that we're still a long way away from having that kind of robust broadband infrastructure in America. And this should frighten anyone who cares about the future of this country as the next generation of Internet apps will require networks capable of sustaining bandwidth.
So if we want to be a global leader in 21st century Internet innovation, we're going to need leadership to come from somewhere on making sure that our infrastructure is delivering the capacity needed to support this kind of innovation.