Backbone IP network Economics - peering and transit

Peering? Who needs peering if transit can be had for $20 per megabit per second?

I last had a detailed look at peering and transit economics in the summer of 2002. It is pretty amazing to see what has happened to prices since then. I have a private mail list underway on this subject and published the first part of a two part study on Saturday.

Details available at

Any opinions here on what MCI coming out of bankruptcy will mean? How will once mighty ATT compete? How can it possibly compete if it actually tries to pay the interest on its bonds?

June 2004 COOK Report page one:

Economic Pressure on Long Haul Fiber

Five Years After Bubble Burst Prices Have Plunged Yet Nothing Fundamental Has Been Fixed

Examination of Data Network Woes Shows Termite-Riddled Foundation Leading to More Bankruptcies in Absence of Broader Understanding

Tech-Telcom Recovery, or just a pause before the next round of bankruptcies?

In this issue we explain why we believe that competitive fiber backbones have a failing business model that has driven prices below the cost of maintenance and replacement. We point out that the carriers are making up the difference with cost cutting in every way imaginable and by subsidizing the loosing backbones with what profits remain from voice. As the voice profits disappear via government (GIG BE), corporate, and municipal networks that are buying and lighting their own strands of fiber, and hence leaving the PSTN, another round of bankruptcies looks to be inevitable.

As Eli Noam has warned, when the dominoes start to fall again, the US government, rather than let the telecommunications system collapse, will have no choice but to step in and start regulating IP networks. The kind of regulation that is coming will have little to do with encouraging common carriage or wide spread access to broadband. Instead it will have everything to do with consolidating a few remaining survivors and enabling them to pay their bondholders. The scene is not a pretty one and the miserable policy is being made exacerbated by American bankruptcy law that permits bankrupt fiber carrier to reorganize through Chapter 11 rather than insisting on disbandment via Chapter 7.

What is especially pernicious about this situation is that permitting bankrupt carriers such as MCI to reorganize makes a bad situation worse. It does nothing to get rid of the company's massive glut of fiber. Instead, by wiping out its debt, it ensures a cost cutting advantage to the reorganized company. Still, blessed with the same massive amount of fiber it had before, it can gain income by again cutting prices below what its not-yet-bankrupt cousins like ATT and Sprint can afford to sell, if they are to every pay off their bonds. Roxanne Googin, in an interview to be published next month, was scathing in her comments on MCI being allowed by the Bush Justice Department to file Chapter 11 rather than 7 - given the company's acknowledged massive fraud. She added that, given the MCI example, in her opinion any carrier that makes a good faith effort to pay back its bond holders is just plain foolish.

With this issue we turn to the badly broken economics of peering and the attendant backbone business models. After three hours conversation with Farooq Hussain and ninety minutes with Farooq and Roxane Googin we have a clearer picture of where things went wrong beginning with the privatization of the NSF Net backbone in April 1995. The plan when implemented was a reasonable one, but it is fascinating to look back and understand how all the dominoes fell the wrong way. The result made a bad situation steadily worse until the Cable and Wireless bankruptcy of 2002 ripped a major hole in the Tier One hierarchical fa�ade and pointed the way to the technology topology business case issues pointed out by Farooq during our conversation of April 4, 2004.

This fairly short issue lays the foundations for the mail list discussion that began on March 18 and will be published in the next issue in about mid May. In focusing on the woes of the LECs during the past two years, we have overlooked the fact that the carriers are in worse shape and that the telcos are again speeding toward the precipice. Most everything is broken and, one of the frustrating issues, is that it is difficult to get agreement on just where. The phone companies do need to die and be replaced. But a huge problem that remains is that, unlike Japan, and many other countries, the US is in gridlock. For, in the US, the ILECS and IXCs "own" the FCC. There is still enough money left in the industry that government is hobbled by a political unwillingness to let the bond and equity holders take the consequences and get on with life.

Looking for the Big Picture and By Passing the Carrier

Let's look more closely at what is going on. One consequence of the fiber glut has been the emergence of a large number of ways to send one's "bits."

By this we mean that there are:

(a) At least three distinct technologies used in sending bits over glass - each with its own economics. Layer one technology: OBGP and User Controlled Light Paths technologies from Canarie that are just beginning to be used commercially. Layer two technology - namely Gigabit Ethernet and 10 Gig Ethernet over fiber, sometimes using Sonet. But more often not. Layer three: IP often over ATM using MPLS rather than point-to-point links.

(b) At least two different ways of interconnecting to (carrier pops and at IX's) and that these can be used by customers to ratchet prices down further and further

(c) That high speed Ethernet VLAN and routed VPN services can be bought from carriers, they can also be used as ways to bypass carriers. One by pass business model is to set up a high speed VLAN transport service at numerous exchanges and then sell customers routes over that service.

(d) That outright purchase of dark fiber is also among the ways to by pass carriers - ie corporate, research and education and municipal nets are all doing this.

(e) That for municipal nets there is a new wrinkle being used by Terry McGarty of The Merton Group. Here anyone who puts 20% of the cost down and goes through the proper application procedures can borrow 80% from the Rural Utility Service (RUS) of the United States Department of Agriculture. The RUS fund has 1.2 billion dollars still unspoken for. Describing this business model at David Isenburg's WTF meeting on Saturday April 3, Terry explained how he is moving town-by-town through southern New Hampshire. He is applying a new and potentially powerful model of economic-broker-middleman in taking his consultancy's funds (20%) and getting the US government to match the remaining 80 percent. He shows town management (Hannover for example ) what is possible. He then applies for the funds with the cooperation of the town. When the application is approved, both parties then jointly hire the contractor to build the network. I recorded his talk but stating that he is operating in stealth mode, he so far refuses permission to publish it. I assured him that I am a friend of what he is doing. However, it remains to be seen whether he will come to that conclusion. The essence of what he is doing is certainly clearly on the public record.

Meanwhile describes Terry in Merrimac new Hampshire in January 2003 and Municipal%20Broadband%20Networks.pdf is a 2002 paper laying out the Merton Group concept for building municipal networks. Think about this and hear the "great sucking sound." A question occurs. Once you do (d) and (e) have you doomed LECs as well as carriers?

(f) That we can expect to hear an announcement before the end of the summer of a venture that will begin the role out of broadband wireless connectivity for video, audio, and data

(g) That another carrier PSTN by pass is the global GIG BE net being paid for by the US DoD.

(h) That all owners of fiber are so eager to sell access and get some income that they will do darn near anything to enable attachment, and that Level 3 and Global Crossing may find their infrastructure superfluous in parts of Europe and Asia when they run up against the national PTTs. As will Sprint, MCI and ATT

(i) That profits from voice propping up the Swiss cheese edifice of fiber data nets are fast disappearing - that the center of the PSTN is being hollowed out every which way with wireless and cable waiting in the wings. As the wireline PSTN collapses perhaps cellular will be what remains? It would seem that only in this context could Verizon's amazing offer on April 12, to the FCC of five billion dollars for ten kilohertz of PCS spectrum make any sense. See

(j) That the USA is in far worse shape than the rest of the world.

  ~ ~ ~ ~ ~ ~ ~ ~

Economic Pressure on Long Haul Fiber

Five Years After Bubble Burst Prices Have Plunged Yet Nothing Fundamental Has Been Fixed

Examination of Data Network Woes Shows Termite-Riddled Foundation Leading to More Bankruptcies in Absence of Broader Understanding p. 1

Backbone IP Networks: Why the Hierarchical Peering Model Is Broken

Like Electric Grid - No One Wants to Pay for Connecting
as Fiber Glut Overturns Dominant Position of Tier Ones p. 6

Some of the Business Model Issues of Peering and Interconnection Or Why the Government Could Eventually Be Forced to Take Over p. 19

We Are Now a Decade into IP Based Optical telecom without a Viable Business Model -- An Optical Network Designer Looks at the Collective Insanity of the Bubble and Calls for Fresh Assessment p. 23

Dave Hughes Explains Sip, VoIP and FW Dialup in Voice Communication with Nepal over the Internet p. 26

Digital Photography Comes of Age -- Two New O'Reilly Books Reviewed in the Context of New Tools p. 29

Isnt the companies still "doped" from the bubble in 2000-2001? Price

You have three cities. Two 12400 GSRs per city, and OC192 to connect them,
that's a total of 12 $150k ($225k minus rebate) cards and let's say $100k
per router for customer facing interfaces etc (unrealistically low).

If you want to pay this investment back over three years and let's say
you'll push 10gigs of customer paying traffic (because of redundancy etc).
You end up with close to $10 per megabit in just equipment fee to
cisco so you have half of the money left from the initally stated price of
$20 per megabit, this for a small inter-metro network.

Since Cisco basically hasnt lowered the price per megabit on any interface
cards for the GSR platform, it cannot be used apart from doing very long
distance transfer via DWDM where the links are full of revenue-generating
traffic all the time. Juniper is even more expensive.

We like these platforms, they're very stable and well performing, but I
just cannot see where they can be justified investing in at todays megabit