Why is this page text-only?

« Two Good Events Tomorrow On Broadband Stimulus | Main | BTOP Should Be About Stimulating Deployment Not Solving It »

April 14, 2009 12:27 AM

Time Warner's Higher Speeds And Lower Caps Collide

Through all the consternation surrounding Time Warner's new bandwidth caps their limitations are always discussed relative to downloading high quality movies. While this is a practical way of thinking about these caps, there's another angle that's even more striking to consider by comparing these caps against the increased speeds of Time Warner's service.

Let's start with some numbers: 100GB and 50Mbps. The first is their top-level cap, and the second is the fastest speed they offer.

Now let's consider some time in the not too distant future where someone invents an application that requires 50Mbps of constant throughput. Let's put aside the fact that there are no apps like that today, and that cable networks have trouble sustaining their maximum throughput. Instead let's assume an app with these requirements exists.

How long would it take to use up 100GB?

First let's convert GB into Gb by multiplying 100 by 8 (there are 8 bits in a Byte). So we have 800Gb.

Now divide 800Gb by 50Mb and you see that it'd take 16,000 seconds of data transfer at 50Mbps to hit the 800Gb cap.

Now divide 16,000 by 60 to get 266.67 minutes. Then divide that by 60 again to get a bit under 4.5 hours.

So tying this all back together, if you were to turn on this 50Mbps application, you'd eat through your monthly allotment of bandwidth in less than 4.5 hours.

This is an absurdly short amount of time.

Even if you assume that there isn't an app that requires a constant 50Mbps but instead that a home would need that much capacity for only 10 minutes a day you still wouldn't get to the end of the month without going over the cap.

I should say at this time that I'm not necessarily against bandwidth caps. I know a number of fiber operators who think that caps are essential to protecting their ability to deliver quality service and to insuring they can afford to pay off their networks. Some of these are even municipal networks so there's no profit-motive driving these beliefs.

But there's an inherent tension that exists between introducing caps at the same time you're increasing speeds. It'd be like introducing cars that can go 200MPH but only for 200 miles. Sure it's faster but you can't really use all that speed.

It's worth noting that other network operators have set higher caps, like Comcast at 250GB, but even these will be overwhelmed as faster networks beget more bandwidth-intensive applications.

Unfortunately our current broadband paradigm doesn't necessarily have an answer for this dilemma. Network operators either need to cap everyone, rate limit the heavy downloaders, or have the cost of their usage eat into their profitability.

The only clear path forward I see to this problem is to push forward aggressively with deploying more networks with more capacity and lower interconnection fees, which ultimately means full fiber networks peered up with each other.

Because without a whole lot more capacity we'll still be stuck in an era of bandwidth scarcity, which means having to manage a limited resource. But if we can get fiber everywhere and fiber networks peered up and acting as a nationwide LAN, then we can infinitely more bandwidth without it being infinitely more expensive.

The takeaway from this post is not that what Time Warner's doing is wrong and evil but that this all highlights the limitations of the current broadband paradigm, which will only be exasperated over time as networks become faster and applications hungrier for bandwidth. And the only way to overcome this is to push towards the goal of realizing a Full Fiber Nation.

Del.icio.us Digg Yahoo! My Web Seed Newsvine reddit Technorati


TrackBack URL for this entry:

Post a comment

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)