[ NNSquad ] Re: NY Times: People are watching much more online video

["nick hatch" <nicholas.hatch@gmail.com>]

On Fri, Oct 31, 2008 at 10:07 AM, Kriss Andsten <kriss@proceranetworks.com> wrote:

On 31 okt 2008, at 05.27, nick hatch wrote:

I see Comcast's move to cap at 250GB as a sticking their toes in the water, and perhaps trying to set a precident at the same time. I don't see a techincal reason for the caps. They don't even differentiate between outbound/inbound traffic. Considering their network architecture where last-mile egress bandwidth is so scarce, that alone makes the whole move seem like a farce.

I'm not sure that I agree there. I don't think I've seen (which is not a synonym for "I categorically deny that it exists") a single capped service that differentiated between upstream and downstream bandwidth. Why? Support.

That's a good point. Support would be nightmare in that scenario, and I can't think of any networks which do differentiate either. However, I don't think that negates my point.

The economics (supply and demand) of ingress and egress bandwidth on an asymmetric bus topology like DOCSIS are very different than Ethernet and other symmetric links. You see demand for upstream bandwidth higher than supply, which is why Comcast is working feverishly to free up analog TV spectrum, move to DOCSIS 3, and implement heavy-handed shaping. All these efforts work to increase effective supply. Increasing the actual supply of bandwidth necessitates node splits: which are much more expensive. (I've heard a figure of ~$100 per home in the service area, subscribers and nonsubscribers included.)

Similar efforts to shape downstream bandwidth are nil; indeed, Comcast introduced PowerBoost for downloads years before they did the same for uploads. 

The cap doesn't help give them headroom in their free capacity on their most limited resource. So what's the point, if it's not strategic? It certainly doesn't seem operational.

-Nick