A New Era of Customer-Owned Networks

Introduction to the Isenberg Paper

Shumpei KUMON (Executive Director, GLOCOM)

GLOCOM follows the work of David Isenberg very closely, particularly his controversial and path-breaking 1997 description of the "Rise of the Stupid Network" (http://www.rageboy.com/stupidnet.html) and his more recent argument published in his own SmartLetter that a new era of "customer-owned, customer-operated networks" has arrived. Isenberg is a GLOCOM Fellow, a former AT&T engineer, and founder of isen.com. We post his SmartLetter essay on our platform with the author's permission.

The Era of Customer-Owned Networks

David S. Isenberg (isen.com)
(Smart Letter #56, June 7, 2001 Copyright 2001 by David S. Isenberg)

Here's one way to interpret the telecom turmoil in the stock market: The carrier-services era is ending and we're at the beginning of an era of customer-owned networks.

Q: How do you make money with customer-owned networks?
A: The same way you make money with LANs.

Service providers (and the companies that provide their equipment) will need to defend-to-the-death any monopoly status they might now have, or suffer radical metamorphosis, or die -- and perhaps all three.

We're beginning to relate differently to our connectivity. We're beginning to decide ourselves what equipment will light our network, how fast it will run, where the fiber goes, how much route redundancy we want (and are willing to pay for), and what gateways, points of presence and service providers our network will connect to. And just as we own desktops and laptops and palmtops and wrist-tops, soon we will own personal networks and home networks and chunks of metropolitan networks and long-haul networks. When they dig up the street, we'll be able to look into the trench and say, "That's my fiber." We'll look at the spectrum and say, "That's my color." (It's becoming easier to be green.)

The act of digging up the streets is becoming pass, if not obsolete. Corning MCS-Road cable packs up to 144 fibers in a 7mm diameter cable. You "install" it by cutting an 8cm slit in the pavement with a circular saw. You kick in the cable, tamp a rubber strip on top and seal the slit with goo. It is up to eight times faster and five times cheaper per route mile than trenching according to Corning.

"Impossible!" the Bellheads rant. "This cable is so close to the surface that it'll get cut every few months," they say. Why, we'd need protocols designed to deliver reliable service over unreliable physical media! (Hint: begins with I.) Plus we'd need physical routing redundancy (and automatic fail-over). So two feeds might only be four times faster and 2.5 times cheaper. I'm crying all the way to the High Bandwidth Bank for Savings.

There are already eight conduits and who-knows-how-much 864-fiber cable buried next to the Garden State Parkway, about a kilometer from isen.com headquarters. Corning's MCS-Road seems like the perfect off-ramp to get that fiber ovah heah.

144 fibers can carry up to 230 terabits per second using today's technology. That's enough throughput for three 64-kilobit voice circuits for every human on earth. (SMART People will check my arithmetic.) But hey, if I had an MCS-Road cable coming down my street, I'd light it cheaply at, say, 100 megabits per second in each direction. Why get greedy when there's more than enough for everybody?

The Corning MCS (Micro-Cabling System) product line expands beyond MCS-Road. It includes two different ways to pull cables through drains and sewers, plus cheap and easy splicing, interconnect and repair systems. Get more details at http://www.corning-cable-systems.de/en/products/mcs I don't understand why Corning is keeping its MCS products such a freakin' secret.

[Note: I have absolutely no business relationship with Corning's MCS unit. My only business tie to Corning is that I was a paid speaker at a Corning CLEC conference this spring. At that conference, I can't remember a single mention of MCS from anybody there -- Corning employee or otherwise -- except me.]

Even traditional, trenched, underground urban fiber construction costs are falling below $1000 per fiber/mile. (Note: not a route-mile or a cable-mile, but a fiber-mile.) Some people say **WAY** below, but let's use the higher figure. Fiber is a 20-year asset, so figure $5.00 per month per mile of dedicated fiber -- cost. Price, of course, would be set in the marketplace (if there's sufficient competition to call it a marketplace).

Some people who know more than I do say that fiber is but 5% of the cost of access. OK, so maybe it costs $100/month to get a lit, connected fiber into my home. If they charge $200, that'd be a good profit even by telco standards.

Some of the non-fiber costs of connectivity include lasers, routers, and radios. Such hardware scales steeply with time and volume. Yet other costs, such as real estate, building entry, reliable power, Internet connections, and networking expertise - especially networking expertise -- don't scale as steeply, but there are economies of scale nevertheless.

[Wait! Here comes the Verizon FTTH crew now!!! Sure, boys, put my 100 megabit jack right there next to my 802.11 hub. Nah, I am hallucinating again. Who slipped the Love Drug into my Kool-Aid?]

Verizon isn't going to bring fiber to my house. The ILECs will destroy their core business with fiber. Newer companies will do it. Dropping costs will make it inevitable. Assuming a free market, of course.

The cost of expertise is falling slowest. Technologies of network simplicity are the antidote. My flying instructor taught me, "A superior pilot uses his superior judgment so that his superior flying skills are never needed." By analogy, superior networking experts will use their superior network design skills to design networks that don't need superior network engineers to build and operate them.

So I think that technologies of network simplicity will advance. Even the Incumbent Local Exchange Carriers (ILECs) want to lower their operations costs. (Whether they'll do what it takes is anotherquestion.)

Local Area Networks point the way towards radical simplicity. LANs were not designed for esoterically trained telco technicians to optimize scarce infrastructure. Instead, LANs were designed for computer owners to hook up their equipment easily, by themselves. As a result, today the LAN, specifically Ethernet, is the front-runner connectivity model for customer-owned networks.

LANs are extending their reach beyond buildings, beyond neighborhoods and campuses, beyond cities. They are driving erosion of the telecom priesthood's prerogative to charge scarcity-based rates for mysterious, arcane services.

Telecommunications carriers that can learn a new diet will survive. Carnivorous high-fat diets of vertically integrated services will have to yield to the lean cuisine of pure connectivity.

Today's stock market debacle is driven, in part, by genuine deflation in the telecom sector. We're doing more with less. Telco stocks stayed too high for too long given the free-fall of telecom infrastructure prices. Now the air is escaping from the long-haul service balloon. It is five years after Qwest laid the first super-abundant trans-continental fiber-optic route, and four years after The_Death_of_Distance (Frances Cairncross, Harvard Business School, Cambridge MA, 1997) appeared.

Ultimately, all connectivity is local. The LAN is pushing out from the building to the metropolitan area. This will take time, but the LAN's disruptive expansion will hollow out and then puncture the ILEC's business model.

ILECs will not give up power lightly. They will lobby for new laws. They are experts at using the courts, and state public utility commissions, and the FCC to further their agenda. Their imperative is to delay or derail the commoditization of connectivity. If they succeed, the momentum of the inexorable Communications Revolution will shift to other countries where public policy is more sharply focused on the linkage of communications and economy. If this happens, the U.S. economy will not be a pretty picture.

Where there is fiber in the metro area, there is perforce way too much fiber. (Everybody buries more fiber than they need -- when you can buy a 1000x option on future growth for just a few percent of the cost of construction, you do it.) Once the fiber is deployed, its owners are motivated to turn it into a performing asset. This puts more fiber on the market, which drives prices down further.

It is already practical and cost-effective for organizations (e.g., the New York Public Library) to own their own fiber-based communications networks.
Large and small businesses will follow. Before too long it will be practical for an individual in a fibered-up city to say, "Give me a fiber (or a wavelength) to my ISP, my telco gateway, my bank, my employer, the movie exchange, and the points of presence for my three closest friends.

Meanwhile, where there's no fiber, the cost of connectivity will remain relatively high. Here, customer-owned wireless router-radios, a la Rooftop Networks (see "Self-Organizing Network -- SMART Letter #28, http://isen.com/archives/991008), form the network infrastructure. These devices are both access devices and network infrastructure. Most of the time they're relaying packets for somebody else. As Tim Shepard pointed out in his famous MIT Ph.D. thesis, as the radios get closer to each other, transmit power drops, receive sensitivity improves, overall throughput goes up, redundancy (reliability) goes up, and all the elements get cheaper. It's a kind of Moore's Law of unwired space. Rooftop (now Nokia RoofTop) is still a player here, along with startups like SkyPilot and UltraDevices. I'm on the advisory board of UltraDevices because I am optimistic about the way the unlicensed unwired space is supporting innovative commoditized network infrastructure. (3G -- whazzat? My 2G phone can hardly do voice.)

Customer-owned networks will bring business changes that won't be easy to understand. We saw the PC getting cheaper and more capable in the 1980s, but only Microsoft understood that the commoditization of hardware meant a value shift to software.

Now we see the commoditization of connectivity. We do not know who will benefit or how, though we can make some educated guesses. I'm guessing that the value of the network will accrue to purveyors of high-value services at its edge. Can you spell MICROSOFT?

Nor do we know who will pay for this infrastructure. We know (and the markets reflect) that telcos are having a hard time of it. Notwithstanding, we know that broadband connectivity to internetworking is massively useful. If we need it (and the technology exists to build it) it will come, just as sewers and running water and paved streets arrived. Furthermore, maybe it won't come to the United States first -- especially if United States policy does not give it priority.

Paul Saffo, the futurist, (there are not many people that deserve this title) says that we tend to overestimate the short-term impact of technological change and underestimate its long-term impact. In a decade, I suspect that we will look back and wonder what took us so long.