Some truth about Comcast - WikiLeaks style

Owen DeLong owen at
Mon Dec 20 15:07:53 CST 2010

On Dec 20, 2010, at 12:16 PM, Rettke, Brian wrote:

> So, we seem to circle the same points:
> 1. Who pays for the infrastructure to support the increased bandwidth requirements?
>        Comcast and most ISPs want the content provider to do so, since they are collecting fees for the service and they are not, but still have to pay for the bandwidth (maintenance and upgrades).
		What do you mean they are not? I'm paying Comcast $100/month to deliver the internet content I want to my home. They damn well are getting paid to do so.

>        The customer is being billed twice: First to get access to the Internet to reach services, and then for the service offered by the content provider. The concern is that all customers, regardless of the services they select, will end up paying for the upgrades if the ISP has to/does raise rates. This makes the customer using the bandwidth-intensive application happy, and the other customers not using it unhappy.
		I don't use particularly bandwidth-intensive applications. However, I do think that access networks should cover the costs of delivering the content I request from the fees I pay.
All that happens if you let them bill the content provider and double-dip is that the content provider has to pass those fees on to the service I'm using (at a markup, of course) who then
passes the cost on to me (again at a markup). I'd much rather pay the cost directly to my access provider without the double (or more) markups, thank you.

>        The content provider pays for Internet access, and in some cases puts in proxies to cache closer to the source. They do not pay the end customer ISP for service (assuming different providers in play). The content provider receives revenue for its services.
		IMHO, this is as it should be.

> The problem is still that someone has to support and build infrastructure. Some believe that Internet streaming video is the direction we are headed in, and that does appear to be true. But there are still a lot of customers that are not using this service, effectively subsidizing the customers using this service. This can be irksome, because most customers are unwilling to go back to a "pay for what you use plan" after having unlimited access. I think that would really put the pressure on both customers and content providers alike to be more efficient.
		If you don't need broadband, subscribe to narrow-band services. They are still available in most areas for less than broadband.

> I understand that the goal is for the customer to get what they want on demand, but that will never be a reality, for anyone, anywhere. I'd love to see content providers continue the push towards more efficient technologies and architecture, but there is no impetus for them to do so unless they have a financial reason. The same is true for the ISP and the customer.
		There are already good incentives for the content provider. It's called "user experience". If the content is close, i get a good user experience. If it is far away, I get a poor
user experience and I move on to a different content provider. If there were meaningful competition in the access market, I could do the same thing. Unfortunately, there is not
where I live and not in most locations.

> Bottom line:
> Customers need to think about the purchase of content (considering each one as a transaction that has value) more. Not as a worrisome, "bill will be enormous" way, but assigning value to it nonetheless.
I think I do this already.

> Content Providers need to continue upgrading methodologies, compression, and technologies in order to make their service a smooth, efficient "essential object." This will help keep any one service from overwhelming the rest, which is the bane of every service provider/transit provider.
I think that is already happening and will continue to happen.

> Service/Transit Providers need to re-evaluate their bandwidth offerings to customers, their relationships with content providers, and with each other. The model is very inefficient and political. The only way to be competitive seems to be, as someone said, to provide a solid Layer 1-3 platform that will drive innovation at layers 4-7.
I think you need to separate Transit Providers from Access Providers here. The reality is that there are four classes of players present without clear delineation:

	Content Providers (including Content Provider Hosting Networks)
	Content Delivery Networks
	Transit Networks
	Access Networks

If there are any pure players in any one space above left, I would be surprised, but, each of these four
spaces comes with a different set of tradeoffs and desires. Traditionally, Level3 has been a Transit
Network with some Access and some Content Provider aspects. Now they are adding Content

Traditionally, Comcast has been a pure Access Network with some Transit. Now they are
adding more Transit and also doing Content Provider things.

As the lines blur, it's going to become increasingly more difficult to define non-peers among these
networks. Frankly, IMHO, the right answer is to stop doing so. Recoup your costs from your customers
and recognize that whatever packets {enter/exit} your network {to/from} a customer, most likely the
other {entry/exit} is NOT a customer, but, a peer.

I know that perspective is probably very unpopular, especially among Access Networks, but,
I think it is the right approach overall. I also think that it is where the market would drive things
if we had actual competition for access services.


More information about the NANOG mailing list