AT&T Explains Its Optical Network Architecture

In Search of Business Models for a Slower Growth Era, Carriers Try to Cope with Hang Over of Fiber and Debt after the Great Optical Build out of the 90's

pp. 1-11 Find out how to order single copy ($125) or group license ($250) for just THIS issue.

We interview John Strand, and Tom Afferton of AT&T. Afferton directs AT&T's Advanced Transport Technology and Architecture Planning. Strand is a consultant for the Optical Networks Research Department at AT&T. He also is a Working group Chair in the Optical Internetworking Forum. The interview describes the choices faced by AT&T in constructing an intelligent optical transport network to take advantage of its new backbone overbuild and the developments of the past five years in optical transport technology.

The amount of fiber available to carriers has exploded as the result of builds carried out largely by new players since 1996. As John Strand points out, by 1995 the optical transport networks of AT&T, Sprint, and MCI accounted for 75% of total inter city fiber deployment in the US. By the year 2000, MCI, Sprint and AT&T fiber had shrunk to less than 1/3 of the national total with the remainder belonging to no less than 39 new national carriers. And, as the year 2000 began, there was an estimated total of 400,000 route miles of fiber with an average 46 strands of fiber per cable.

One may argue that one possible reason for this build out and the consequent huge amount of debt taken on by the telecommunications industry is a misunderstanding of the growth rates of internet traffic. We cite in detail the argument of Andrew Odlyzko that although for a time in 1995 and 1996 internet traffic likely did double every 90 to 120 days by 1997 growth had slowed down to about 100%a year. The consequences of a difference in growth of between 800 to 1500 percent a year and 100% per year are now being felt both by the owners of the transport and builders of the optronics and routers. See .

Odlyzko presents a well-supported series of arguments as to why the doubling of growth three or even fours times a year was a myth. It was nevertheless a myth that shaped a stampede that led to the laying of vast amounts of fiber by companies seeking to win what was an impossible to maintain sprint of growth and obtain a first mover advantage where those who lagged behind would never be able to catch up to the one or two players that quickly became dominant.

Optical technologies did yield 100 times and more recently probably even 1000 times advantages in the cost of moving bits in an abstract sense when compared to the legacy networks of ATT, Sprint, WorldCom and the ILECs. A first mover advantage there could overturn global telecommunications. There were, however, only two problems. First the amount of capital required to rebuild the infrastructure of the global PSTN was simply too vast for any single company no matter how ambitious its borrowing and spending plans. Second, as the largest carriers moved forward to try to win the technology transition race, they moved to buy single-vendor optronics solutions which were the only thing that could be quickly rushed into service.

The development of optical internetworking standards was too often over looked in the drive to buy whatever solution seemed to fulfill the promise of capturing first mover advantage. Industry players did found the Optical Internetworking Forum in 1998. As we learn from our interview with Strand and Afferton, the work of the OIF will play the critical role in deciding how soon the new fiber networks will be able to really inter-operate and, in like manner, how soon they can rely on an open national market that uses their infrastructure to sell bandwidth rather than rely on what ever small part of national or global demand they can meet on their own.

Meanwhile. AT&T, like many other traditional carriers is faced, with the investment of very large sums of money necessary to make the new optical technology a part of its infrastructure and prepare for the day when the voice traffic that accounts for only about half the total but brings in about 6 to 7 times as much money as data traffic shrinks to very small minority of total traffic. The problem is of course made more difficult by the fact what is a cost efficient architecture for circuit switched voice is not one for data.

As gigabit and ten gigabit Ethernet begins to blur the difference between LAN and WAN, enterprise customers increasingly will want bandwidth that runs seamlessly from an office in New York to an office in California. One of the major concerns for AT&T, which is rich both in metro area fiber and inter city fiber, is to architect a network where the differing requirements of long haul and metro optronics will not be a barrier to the rapid provisioning of customer circuits. AT&T and the other carriers are well aware of the concerns expressed by Avi Freedman in the May COOK Report that provisioning such circuits now takes months.

According to Afferton, a key feature of what AT&T is building relies on "SONET multiplexers that have added intelligence and can handle not only SONET TDM type traffic but also, for example, can map Ethernet into SONET, and can do some ATM switching and so forth. We have started rolling these out into our metro networks in order to reduce our costs, as far as SONET traffic is concerned, but also to have the intelligence to inter-operate with the intelligent optical switches so that we can then do this automated provisioning end-to-end."

Rapid provisioning and ultimately giving the customer the opportunity to do the provisioning is the goal by year's end. The current economics of the technology transition dictate that any carrier with a large amount of fiber will be selling what ever flavor of bits including lambdas it can find customers for. AT&T has been selling lambdas since 1999 as it continues to upgrade its network with switches from Lucent and NEC.

Level 3 Wants to Be Global Carrier's Carrier

An Introduction to its Structure and Lines of Business

pp. 12-17

We interview Andrew Morley, Senior Vice President of Global Strategy at Level 3. Morley recounts Level 3's impressive global infrastructure and lines of business. One model for survival in the industry's current debt crunch is for a company to build a global green field infrastructure so substantial and cost effective that it can convince the traditional carriers that it is more cost effective to outsource new data and perhaps even voice services to Level 3 than to invest the huge amounts of money necessary to upgrade their own infrastructure further.

Based on additional interview that we have done with Ron Vidal and Rob Hagens at Level 3 there is evidence that Levels 3's preferred way out of the present situation is to adopt a kind of horizontal business model as infrastructure supplier to legacy vertically integrated carriers. In any case it needs to gain as much income from its investments as it possibly can so that it can pay the debt on its bonds before it runs out of cash. Given its resources if it cannot do this, the portent for the future of the industry is especially grim for Level 3 would currently seem to have more going for it than the rest of the new green field players.

Still the need for increased cash flow has resulted in tightened budgets and in a degree of ambiguity about its mission. According to Morley, Level 3 has 3 major lines of business: transport (sale of lambdas and dark fiber), Internet Protocol services and Softswitch services. For 2001 he declared that "Transport should be about 45%-50% of our ongoing communications revenue, Softswitch about 25%-30% and IP services about 20%-25%."

Softswitch is one of L3's biggest success stories with sales of its managed modem Softswitch services running at 8 billion minutes per month. A different source informed us that these services in the past year have for AOL and EarthLink brought the monthly cost of maintaining a dial up port for internet service down from 15 to 3 dollars. To keep its large enterprise customers happy Like AT&T, Level 3 is focusing on rapid provisioning of bandwidth with a claim that its guarantee of less than 30 days from order to turn up of service is the best in the industry.

Cable and Wireless Cuts off Peering with PSI for Five Days

Both Networks Get Low Marks in Nanog Debate on Peering Policy

pp. 18-21

From June 2 through June 5th Cable and Wireless turned off peering with now bankrupt PSI. We publish an edited version of some of the discussion on NANOG which shows that the reputations of both parties were tarnished. Peering is still a matter handled privately and without any concrete standards that offer much assistance to those who want it. Because it is a vehicle without firm rules, most service provides feel that it discriminates against smaller players and leaves the industry prone to significant disruptions in service. peering disputes are one of the few occasions when players will 'vent' enough to give the rest of us an idea how these critical relationships affect the rest of the industry.

Security Consultant Chronicles DOS Attack

p. 21

Steve Gibson, Gibson Research Corporation has published on his website: The Strange Tale of the Denial of Service Attacks Against "Nothing more than the whim of a 13-year old hacker is required to knock any user, site, or server right off the Internet.

I believe you will be as fascinated and concerned as I am by the findings of my post-attack forensic analysis, and the results of my subsequent infiltration into the networks and technologies being used by some of the Internet's most active hackers." Gibson ends his essay: We need a tool to hold ISP's accountable and publicly demonstrate individual ISP irresponsibility. Given the universal reluctance they have demonstrated so far, I believe that only active public scrutiny will bring about the changes required to insure a reliable and secure future for the Internet. The development of that tool is my next project. The name of that FREE tool will be: Spoofarino tm"

Ex Employee Files $150 Million Lawsuit Against NSI Alleging Knowledge of Violations of ICANN Agreement and Other Wrong Doing

NSI Closes Domain Policy List, Mueller Critiques ICANN's Position on Root

pp. 22-25, 30

NSI hit with $150 million lawsuit. We summarize the events of the release of Michael Johnson¹s allegations against NSI ranging from "not fully responding to the request from the National Telecommunications Information Agency when it expressed written concern about security issues on the server, at a time when the officers of Verisign knew from its customer¹s and internal technical management team of serious incursions from outside sources because of the system¹s security weakness" to "personal knowledge of instances of violations of the ICANN Agreement, predatory practices against competitors, disregarding the law when compliance therewith should stop wrongful conduct, ignoring conflicts of interest which produce a competitive advantage for Verisign." ICANN was supposed to ride herd on NSI. Instead in the absence of any oversight ICANN stands by and collects its tithe from NSI which, having just been given an essential permanent monopoly over dot com without warning shut down the mail list on which the Johnson allegations were published. We would like to move to a competing registrar, but having seen the trouble that NSI's system has created for others (including loss of domain) we would not dare attempt it without the assistance of a consultant. Some competition.

We include several pieces from ICANN watch. Milton Mueller¹s critique Are Multiple Roots Outside the Scope of ICANN? Also Registrar War Developing- NSI joins battle

Vint Cerf's Doublespeak on ICANN Elections

As Decision About the at Large Elections Nears ICANN Tries to Find Any Excuse Maintain Grasp of its Insiders

pp. 26 - 27

Make no mistake about it. ICANN will not allow any more open elections for Directors. The new ICANN budget has no line items for an election in the next fiscal year ­ millions for regulatory control and none for open democracy. In mid May ICANN¹s behavior was parodied in a web based cartoon ³movie² at there Vint Cerf is shown as a two dimensional cartoon like puppet carefully following the Jones Day scripts he is given.

During the Stockholm meeting just ended Cerf said: "The idea that a particular infrastructure service, kind of like the power system or the road system, needs to have a kind of global election to manage the underlying infrastructure is hard to fully appreciate." We publish a discussion that shreds Cerf's labored reasoning.