Feed on

Recently a list member quoted Art Brodsky as saying that a major question was that the CWA would be key to “whether the Internet will continue to be open, or whether the telephone and cable companies will turn it into an instrument under their control. The prospects are not encouraging.”

Part of this issue is lurking in the many layers of NDA encrusted peering agreements.

The point is that determining the cost of backbone interconnection is one of the most arcane black arts in the internet today. The commercial internet backbone dates from April 1, 1995 when NSFnet was turned off. UUnet PSI and BBN and MCI were the principal commercial backbones. These guys got together and said lets agreed that since all our networks are of approximately the same size we will interconnect with each other and freely delivery each others traffic. All interconnected networks that can deliver their customers traffic to the entire global internet shall be known as Tier one backbones. If you paid to connect to a tier one backbone - you knew you had a rock solid upstream provider. NO One was upstream of a tier one cause they were at the very top of the pyramid.

The rules were informal.No one entity tried to be number one rule keeper. The art of peering quickly evolved- Tier one peers did not have to deliver pay any one upstream On the other hand every one else was downstream they got money from their downstream isps for links running at a defined speed at such and such cost per month ISPs to whom they were providing internet connectivity. This worked pretty well. Pretty soon ISPs figured out that they could save money by peering with each other. Keep local traffic local. Only traffic that was non local really had be paid for. Delivery of Non local traffic to the rest of the internet was called transit.

Now argument began to develop over transit Exodus in 1997-1999 was a network composed mostly of web farms situated at interexchange points. Exodus said to BBN GTE we have the content for your eyeballs peer with us for free. BBN GTE refused. They won. Exodus did not disconnect. But the lesson was learned.

Large amounts of symmetrical traffic would save money. Tom Vest Built up AOL’s backbone in the pacific region especially by building data centers that changed traffic patterns in the pacific between 2000 and 2003.

Google looked at that and some would copied it with great success. Google’s best keep secret is its huge network. We have descriptions of Google’s large server farms…. what we don’t have and don’t know the the extent of their microsever farm in a sixteen wheeler Truck. How many thousands of these exist?

Has Google been able to turn itself into a tier one network? Quite possibly. EBAY and Amazon have built up global networks of a tier one size. SBC has a service contract with Level 3 for its backbone ie it runs on Level 3’s net.

Now think about the cost of inter connecting to the internet. ALL of this was funded in avery few short years by teen agers who for a few thousand dollar worth of equipment could setup internet service in their parent’s homes. In 1992 Avi Freedman set the basic nationwide rate for dial up service per month. $19.95. By 1997-98 there were 10,000 commercial internet services provider business in the USA. All run on.. shoe strings by hardworking passionate men and woman. There were LOCAL companies who really did care about their customer and who would give them GOOD SERVICE. All for a measly $20 bucks a month. Still if you were at the top of the pile there was lots of money being paid in by the little guys. The big networks still had lots of folk to pay for the backbones they were providing. But with broadband raising the cost of entry circa 2000 onwards and the duopoly driving thousands of ISPs out of business. Twenty bucks a month for comcast or qwest or sbc didn’t go so fart as it used to. The large companies brought dis economies of scale to pay for and operate national back bones.

We have built a superb communications system, the safe operation of which is critical to our economy. Yet because it has grown up with out regulation we actually have NO idea what it does cost to run and with out knowledge of costs we have no way to figure out what prices should be. I met art brodsky when i was at OITA in DC in 1990 1991. He is an outstanding reporter but I have strong doubts about the thesis that the CWA is the focus of our problem,

I’d say to bob it is the interconnection of the first square mile that is the problem.

Later I added: It is possible that Google is sitting there laughing at the need of verizon, qwest and ATT to just deliver the bits without google paying anything if it is a tier one peer. Does Google pay? Something surely but what we just don’t fircking now.

Of course the simple point is that we have NO clue about that global infrastructure at google that Erikas Napjus is director of.

Policy makers are beinbg asked to make critical decisions with no knowledge of specific facts That Is BAD

One Response to “Peering and the Internet Backbone Business Model - Top Secret”

  1. on 02 Nov 2007 at 4:42 am More on peering | Susan Crawford blog

    […] what happens on backbones is not available to us or, more importantly, to researchers. Gordon Cook says that everything about prices for backbone carriage is […]

Trackback URI | Comments RSS

Leave a Reply