Why a Layered Model is the Only Reasonable Way to Evaluate Telecom

Lines of Business Have Blurred - Making the Regulatory Concept of Vertical Silos Archaic

Time Has Come to Bury All “Bellhead versus Nethead” Polemics

Click Here to get the Full Report in .PDF Format


Since the bubble burst in late 2000 sending the Internet and the rest of telecom into a tailspin, it has been rather obvious that former stability and predictability of the economics of telecommunications has been shattered. The last several issues of The COOK Report have explored the fallout of those shattered economics in great detail.

In this introduction The COOK Report notes that telecom economics is likely to stay broken, first, due to oversupply, and second due to lack of differentiation on anything other than price across too many competitors in services and service providers. Third: because of very loosely bonded brand loyalty. A final and very serious additional problem is regulatory instability as the FCC struggles with historical precedent in its interpretation of legislation. It finds itself whip-sawed between its “vertical silo” model derived from the technology it regulates, and the increasingly advocated “horizontal network layer view” of the IP enabled services, including but not limited to VoIP, Video over IP, and so on.

As long-term, and, perhaps, not so long term, readers of The COOK Report are aware, this publication has not only long trumpeted the “bellhead vs. nethead” divide, but taken a partisan position where anything seen to be “bell-headed” was regarded as bad while “netheaded” was seen as the ‘nirvana’ to which the Internet would guide telecommunications.


“Nethead” versus “Bellhead” Is Now "Tired"

“Nethead” versus “Bellhead” is a metaphor that at a moment in time in the mid to late 90s in the pages of Wired magazine was used to hype the new civilization changing Internet. It tried to encapsulate in a few broad brush stokes an Internet way versus a telephone company way of thinking about the changes brought about by the launch of the commercial Internet.

The dichotomy was based on the assumption that the Internet players possessed a level of understanding and ability that escaped the telephone companies. Like many of the assumptions that powered the dot com boom, the net head visions did not give their standard bearers enduring business models. Furthermore as TCP/IP founded technology grew and began to mature, the boundaries of the “nethead versus bellhead” world have now blurred as everyone enters everyone else’s line of business. The metaphor was once powerful as a polemic. However, four years after the bust metaphors used in polemics cloud understanding. If we are to use metaphors in our attempts to understand reality, the metaphors should provide an accurate reflection of that reality. This introduction will explain why “nethead” versus “bellhead” can no longer provide such a reflection and hence should be banished from our thinking.

In parallel with the network economics mail list discussion from May to mid August, I have been, having private discussions with Melissa Davis and Frank Coluccio, two people who are no strangers to readers of these pages and who have been working on enterprise and corporate networks for decades. These two for many weeks have been perplexed at what they refer to as my insistence on carrying on the banner of a “false dichotomy.” Melissa Davis has said time and again that the assumption that there was a de facto dichotomy had made the polemicists blind to the fact that the packet Internet rode on telco and carrier TDM circuits; that ATM and Frame Relay, telco offerings, were, in point of fact, packet networks, albeit at a lower layer than IP. Frank Coluccio described to me in detail the topologies of remote access, Metro access, campus building access. The COOK Report has spent a decade on developing a publication platform devoted to proselytizing the paradigm shift of a stupid, end-to-end internet with independent “smart” publishers at the edge who could say that the titans of the old industry had no clothes. Unfortunately, The COOK Report had become so devoted to defense of the paradigm that it has almost missed very significant on going changes and perhaps never initially understood at an adequate degree of detail the interworking of some of the new technology.

As embodied in the various telecommunications legislation since 1934, wireline telephony has been seen as a stand-alone service all the way from customer handset across physical twisted copper pairs, through a labyrinth of switching to the sound delivered from the customer headset. As developed, wireless telephony had its own vertically integrated structure, as did broadcast and cable TV / cable Internet. In each case everything from the physical plant right up to and including the service delivered by that plant was treated as uniform, legal, economic and regulatory whole.

Similarly. Internet service providers were seen as independent companies that integrated physical facilities using TCP/IP to bring email and then Web and other computer-based services to subscribers. In the current situation, as I talked with Melissa and Frank, seeing the telcos versus the ISPs as vertical integrated ‘silos,’ led me to lump the enterprise in some imaginary silo of its own and, following the lead of some “nethead” commentators, to talk about the enterprise “leaving both the PSTN and the Internet” as it built out its own network services. The “nethead versus bellhead” dichotomy was not helpful because it .focused our way of seeing on those vertical silos, even though the walls separating them were dissolving.

Looking at it Fresh

Not a change in what is being viewed, but a change in how one sees what is being viewed.

Based on those discussions - to be explored further in the December portion of what will become a November - December issue - The COOK Report concludes that there is value in looking how this Third Force, the Enterprise Networks of large corporations and governments, are constructed. If we understand what are the business drivers and constraints, we may then comprehend the interworking of the component pieces of networks in general.

Flying under the radar of the popular “tech-press”, corporate and government Enterprise Networks reveal themselves not to be a new phenomenon at all. Historically, they have been different in internal structure and operation from both the PSTN and the public Internet. While these differences will be further explained the December portion of this November December issue, here I point out just that the differences arise from the very different operational dynamics in their eco-space.

As Frank and Melissa have led me to see: The Enterprise Network gives the lie to a “nethead versus bellhead” view of reality that was never very accurate, and is now rather useless. As Melissa says: “Money does not always tell the truth, but it is an excellent tool to reveal the truth in fact.” She has asked us to observe television commercials and Internet Ads, that are clearly financial investments made with hoped for ROI as a metric. During the 2004 Summer Olympics there were numerous commercials and Internet Ads for VoIP, most of which have originated from AT&T, Verizon, SBC. She asks: “If the count for commercials for circuit-switched voice equals zero, then does that not prove the irrelevance of your “Nethead” vs “Bellhead” dichotomy? Are not the telco’s also ISP’s with their data lines and DSL? When wireless telephony also has Web and email access, cable Internet exists with Vonage over the cable connection, where then are your clearly demarcated silos? Have you been fitting your perception to reality, or, conversely, fitting reality to your pre-conceived perceptions!”

Nor, is analog vs. digital the crowning achievement. As time passed telco lines were digitized with the DS-1/T-1 hierarchy. The only analog in the old circuit switched network is from the end-user handset to the Class 5 switch (analog signal over an analog carrier. RF (wireless) is an analog carrier, as is microwave, and photonics.

Through the Glass Darkly

In the midst of all this, the layered network model, well familiar to those in data networking, was implanted over the past 30 years. It was a work of superb engineering brilliance that enabled the drivers that enabled networking to explode. Networks would adopt layered models. The FCC Computer One Two and Three Rulings of 1971, 1975, and 1985 enabled unregulated data networking by recognizing for the first time the concept of layers (alongside the silos) starting with basic transport (regulated) at the bottom and adding enhanced services (unregulated) on top.
From the point of view of the technology, each network layer would be functionally independent of the ones above and below it, acting only as server to the ones above, and client to the ones below. Each layer only communicated with the corresponding layer of the network stack on the receiving or destination end-machine.

As Richard Whitt says on page 8 of the MCI paper cited below:

“Functions are allocated to different protocol layers or levels, with standardized interfaces between layers. The flexibility offered through the layering approach allows products and services to evolve, by accommodating changes made at the appropriate layer rather than having to rework the entire set of protocols. In other words, layering allows changes to implementation of one layer without affecting others, as long as the interfaces between the layers remain constant.”

To steal a quote from Michael Powell below, this in turn transforms modern digital networks into a metaphorical equivalent of “Lego blocs” that can be plugged together in various ways. Networks can, because of the standardized interfaces between layers, plug in a video lego block if they choose. Or they may add or subtract various combinations of various audio application blocks. With this approach of building bricks and layers the concept that things must be built from the ground up in one standard way disappears. You can snap your Lego bricks together in a myriad of ways. When you realize this, you also realize that the bell headed, voice, phone company silo is no more.

Things are converging but not toward any uniform box that does it all. They are converging because in addition to the standardized and uniform way that applications communicate between layers, there is now a predominant set of single protocols at the logical layer - namely TCP/IP - that can interact with almost any application and its content from the layers above and encapsulate everything that it is offered in a standards-based manner and place it into a wide variety of data link protocols and physical transport devices below.

In other words more and more informed observers are beginning to say that the viewing of horizontal layers is the more appropriate market economic analysis tool for understanding what is happening as everyone gets into everyone else’s here-to-fore vertical business.

Permeable Layers Replace Both Bellhead and Nethead

In March, 2004, MCI published a whitepaper - A HORIZONTAL LEAP FORWARD: Formulating A New Public Policy Framework Based On The Network Layers Model, by Richard S. Whitt, Senior Director of Global Policy and Planning, MCI. I will say more about this paper below. Among our discussants Melissa Davis finds much to recommend a layered view. Andrew Odlyzko stated in this issue’s mail list discussion “I actually think the layer model is the natural way to go with regulation, it’s just that I do see difficulties in implementing it.” As Whitt points out, in the last two years Lawrence B. Solum and Minn Chung, Douglas C. Sicker, and Kevin Werbach are among those writing scholarly papers advocating a layered approach to telecom regulation and economic analysis.

Finally, although we have taken the FCC’s Michael Powell sharply to task for his views that seem to favor the LECs, if one looks at ongoing reality from the layered point of view, one certainly sees the LECs, themselves, striving to change by adopting the Lego blocs necessary to deliver whatever combination of digital voice, video and internet services they can most cost-effectively build on the infrastructure they own.

Moreover, Powell himself in a significant but little quoted interview with the Gartner Group on June 14 2004 sees things increasingly in layered terms.

Gartner Fellow Ken McGee: “You’ve stated many times that broadband deployment is your central communications policy. Why is that so important? And what are the benefits that we’ll derive from it?”

Powell: [mentions history of the decision before World War II to grant ATT a regulated monopoly and then says]

“So as I look at transitioning to the communication platforms of the future, I see that the beauty of Internet protocols is you get the separation of the layers between service and technology. Unlike the phone system, which is engineered around an application, the Internet layered model allows you to, in essence, separate applications from infrastructure. What’s exciting about that is almost any platform can be a broadband bit-carrying platform.” [snip]

“ I’ve started thinking about networks as Legos. There are red bricks and blue bricks, and you can snap the bricks together. Our notion currently is the network, this network and that network. Networks are going to be separable and distinguishable from the way we think of them today.” [Snip]

McGee: “Why do we need an AT&T, MCI or Sprint anymore? Would you agree with the scenario that it’s over for the inter-exchange carrier?”

Powell: “I think you have to be in the communications business. And for every segment that you’re limiting yourself to, you’re decreasing your viability. So, no, I don’t think there’s a compelling case for a stand-alone long distance company over a long period of time. . . . We know what the cost of a long distant minute is - basically nothing. You can sell nothing for a mark-up for a while, but only until something starts eating away at it.” [snip]

“There’ll be room for small companies and there will be niches. But it’s all about access to an infinite amount of information, and whoever can get organized and present that information or data - whether news, information, entertainment or basic communications - is going to prosper. And anybody who thinks in narrow platform terms such as cable, satellite or wireless is going to lose; they’re not going to get it fast enough.”

Editor: Powell is certainly saying that the time for segmented vertical silos is fast running out. The entire interview available at

Struggling Toward New Insight

When I started privately debating this with Melissa and Frank almost a month ago, my initial frame of mind was to see the enterprise as leaving the PSTN or even the public Internet and heading off to build its own enterprise vertical silo. But was this really happening? I now think not. Why?

Because, as Melissa points out, that isn’t the way Enterprises work, nor the way they do business. They are conservative from planning to finance to engineering. They do not “build” anything new unless it is the result of an analytical “make-or-buy” decision. Enterprises don’t say: “Melissa, just go get us a pipe from x-to-y, integrate with the other letters of the alphabet, and get the biggest fastest pipe available.” Enterprise IT knows it isn’t that simple, that changing bandwidth absent business requirements increases costs for engineering, the pipe, re-provisioning, changing the CPE, perhaps the routing tables, the network monitoring baselines and thresholds, perhaps the capacity of the other gear behind the Edge router, the interface on which it connects, the interfaces all the way across the geographical distance. Vendors may draw a straight line from LA to NYC, but that is a huge leap of abstraction, masking a great deal of work and expense that the bandwidth hungry never see because the Network Layers hide that from them so they won’t have to know about it.

Looking at the enterprise is important because that is the one place above all to get an adequate understanding of how networks are built from many different pieces for many different purposes. The Enterprise reality shows us that the only way to build a network is by taking, as best one can, from a complicated set of choices that are specified not by a set of ideological goals but by a unique set of customer needs as iterated by the network architect in a process designed to meet application and customer requirements while preserving as much future flexibility as possible. “In the corporate IT world, the half-life of those who make decisions based on ideology is pretty short. Network decisions are made on parameters of robustness, stability, availability, redundancy and fail-over, reliability, vendor support, cost of change, maintenance histories, extensibility, and price. Look at the difference in that matrix and the vastly more simple way a residential Internet user makes network equipment and vendor decisions.”


Why a Layered Model is the Only Reasonable Way to Evaluate Telecom

Lines of Business Have Blurred - Making the Regulatory Concept of Vertical Silos Archaic -- Time Has Come to Bury All “Bellhead versus Nethead” Polemics p. 1

Building a Post Crash Greenfield Carrier:

The Development of Reliance Infocomm, Inc. (India) and Reliance Communications Inc (USA) -- Build in India - Buy Access in US with Caution and Few People p. 8

Cash Flow Evaluation of Medium Size Service Providers

Evaluating Market Strategies for Survival Among Estimated 200 ISPs with 5 to 100 Million in Annual Revenues p. 13


Symposium Discussion -

Telecom Technology and Economics Dissolving Operational Boundaries

PSTN versus Internet Dichotomy No Longer Useful as Operators Construct Varied Networks to Suit Varied Customers p. 21

Why Fiber to the Curb? p. 21
Quality of Service in a Best Effort Network – High Stakes Arbitrage? p. 22
Years Later – Still Struggling with Packets in TDM Networks p. 23
The Enterprise Takes More Voice off the PSTN p. 24
Boucher VoIP Deregulatory Measure and Potentially Ominous Uses for ATM p. 25
What is the Current Role for ATM? p. 25
Where MPLS May be Headed in 2005 p. 27
ATT Leaves Consumers Behind p. 29
Telco’s and Sales p. 31

Be Careful How You Think About “Legacy” Systems p. 32
What Does the Enterprise Need to Know to Make Intelligent Network Build Decisions? P. 33
The Fate of the PSTN - Likely to Morph - Not Collapse p. 36
So What is Really Happening with VoIP? p. 37
Packet + Switched = HPSN p. 38 - Bride or Bridesmaid? AT&T and MCI May Compete for Suitors p. 39
Voice Costs and the “DS0 Mile” p. 40
How the Delivery of Voice Will Evolve p. 41
MCI’s Layered Regulatory Paper p. 42
Economic Benefits of Networks p. 42
Backbone Traffic p. 45
What About Sonet? p. 45
Side Bar:
A Late Summer Snapshot on Transit Prices from Nanog p. 30

Contributors to this Issue

Affiliation given for purposes of identification - views expressed are those of the contributors alone

Jim Baller, Partner Baller Herbst Law Group
Peter Cohen, Peering Director Telia-sonera
Frank Coluccio, President of DTI Consulting and designer of optical networks in New York City
Melissa Davis, optical network architect formerly with Cisco and now with RS Information Systems
Beth Gage, Manager, Solutions Marketing, Juniper Networks
Peter Juffernholz, former Peering Director Teleglobe, currently Developer of Business Alliances DT
Ram Krishnan, Senior Director Technology, Axiowave Networks
Pete Kruckenberg, Architect Utah Education Network
Francois Menard, Canadian policy expert and municipal fiber network architect
Andrew Odlyzko, Director Digital Technology Center, University of Minnesota
Dave O'Leary, Architect Juniper Networks
Jere Retzer Sr Mgr, Next Generation Networks, Oregon Health & Science University
Larry Roberts, Arpanet pioneer and CEO Caspian Networks
Dave Siegel, VP Network Architecture & Long Range Planning, Global Crossing
David Sandel, Chief Technology Officer NetLabs LLC
Darin Wayrynen, CTO GoodNet, Former VP Enginering Winstar, current ISP operator
Matt Wegner, Product Manager North America, Packet Front
Damien Wetzel, Network Consultant paris, Formerly with Akamai and Internap
Ronald Yokubaitis, CEO Giganews