Print 

Tools for Access and Scaling: Ethernet in the First Mile, 10 Gig in Backbone, ENUM In PSTN

Jonathan Thatcher & Howard Frazier Explain Standards Goals Of EFM,

Discuss How Ethernet Is Changing the Access Space Impacting Product Development, Time Lines, & Broadband Infrastructure,

pp. 1 -23 Find out how to order single copy ($125) or group license ($250) for just THIS issue.

We interview Jonathan Thatcher, principal Engineer, World Wide Packets and Chair of the IEEE P802.3ae, 10 Gigabit Ethernet Task Force and Howard Frazier, CTO Dominet Systems and Chair of the IEEE Ethernet in the First Mile Working Group. We have woven the two interviews together to tell a comprehensive story beginning with Reasons for Use of Ethernet in the Access Space; Ethernet in the First Mile standards development; Standards Impact on Product Design and manufacturing; Impact of the Copper Specification; Ten Gigabit Ethernet Standards and Product Development; Impact on SONET; and ending with Global Impact

Part One: [Thatcher] Technology Issues in Broadband Access Meeting the Needs of the Access Space What Differentiates Ethernet in the Access Space

Part Two: [Frazier] Focus of the Standards Group Architectural Issues OAMP Issues The Copper Media Specification - 3 Proposals Basic Elastic Networks The VDSL Standards

Part Three: [Thatcher] Approaches to Standards Development The Standards Process; Goals and Strategies Standards Issues to Consider When Developing a Product Standards Issues in Manufacturing Other Impacts of Standards

Part Four: [Frazier] The Copper Specification's Impacts Impacts on the ILEC Impact on Cable Access and Project Pronto Timeline for the Standards Effort Assessing the Market Place Dominet

Part Five: [Thatcher and COOK Report] Ten Gigabit Development Ten Gig Impact on Backbone Is SONET Dead? [COOK Report] Impact on Network Topology View of Overall Marketplace

Part Six: [Frazier] Global Perspective - Ethernet and IP as a Utility

Some excerpts from the 22,000 word discussion: To answer the question of why Ethernet in the access space, Thatcher points out: "Ethernet is simple, flexible, has proved extensible, easy to manage, inexpensive to operate, and has a good reputation for interoperability. But there is another reason why Ethernet is so important here. We are seeing consolidation at each layer of the OSI stack, starting at the bottom and working up, into fewer and fewer technologies. Ultimately, this will give way to a single choice. For anyone wanting to build access network infrastructure, Ethernet is now really at the top of the list as one of the things that you have to decide to incorporate. There are no other options. It is the protocol, which will endure."

"WWP is not tied to the choice of one media over another. We do recommend point-to-point fiber as the only topology that can allow greatest flexibility for upgrade or future-proofing. Outside the building, anything over copper clearly has a limited life. A point to multi-point fiber infrastructure makes it extremely difficult to upgrade one customer at a time. It also has all the normal drawbacks of a shared media. We will give customers Ethernet over whatever media they want. But we do business in only one protocol and that is Ethernet."

Howard Frazier's outlook is a bit different. He describes the scope of the EFM development effort. There are four areas of development. Three of them focus on the physical medium and bit level of transport. Those are Ethernet over ordinary voice grade POTS wiring. The Standards Committee hasn't said anything more specific at this stage than Ethernet over copper. But since the Ethernet standards already have lots of specifications for running Ethernet over copper, it implies that we are working on Ethernet over some new form of copper that we haven't applied ourselves to before. This new medium may be a copper twisted pair of Plain Old Telephone Service wiring run over considerably longer distances than we currently support on the copper medium."

"Another area is Ethernet over point-to-point fiber. Again there are plenty of standards for Ethernet over fiber optic cable, but what is new here is that you are talking about Ethernet over a single fiber optic cable where you simultaneously transmit and receive over one strand of fiber. We have not done that in Ethernet standards of the past."

There is a third area of technical concentration on what we called passive optical networks. This will use a point-to-multi point fiber optic technology where you have a single fiber coming into the Central office, or perhaps a pair of fibers-and then through passive optical components-splinters, combiners, perhaps wavelength division multiplexors, you fan out to individual subscriber sites. In this case, the point is that the equipment deployed outside the Central Office and into the outside plant environment would be passive optical components. In other words, there would be no active electronics."

"There is also a fourth area of work on operations, administration, maintenance, and provisioning protocol's, (OAM and P) - management tools, software and support. This OAMP is obviously of great concern to service providers because they want to be able to maintain, diagnose and operate their network and be able to use familiar tools and mechanisms as they do so. Consequently one of the things we have to do in the Ethernet in the First Mile group is build some OAM and P features into Ethernet and, for some of the existing features that we have, export them in a way that is familiar to the service provider."

Frazier: "My personal opinion is that Ethernet over copper has a very large market opportunity. Why because there are lots of people who have copper wire coming into their home or small business. I like going after technology with estimates of 40 million U.S. customers and a hundred million worldwide. Those are good numbers. I think that your assumptions are probably correct. But it's not an exclusive thing. You cannot just look at Ethernet over copper and say that is your solution. You have to look at Ethernet over fiber as well. Otherwise you don't have a product to offer."

On ten gigabit Ethernet Thatcher says "we know at this point exactly what the standard is. This means we know exactly how a product that is a 10 Gigabit product should behave, how it should function and what characteristics it should have." "My personal belief is that 10 Gigabit Ethernet is a bit behind where one Gigabit Ethernet was from a market development perspective at the same relative point in time. At this point in gigabit development we were seeing all sorts of products which claimed to be gigabit compliant and were certainly working, although not all were interoperating. In 10 Gigabit we're not quite there yet. I think that the major reason for this is the optics. In the case of one Gigabit Ethernet we used optics which were virtual clones of what was in gigabit Fiber Channel." "Before the one Gigabit standard even started to be written, we already had a supply of volume optics that worked at a gigabit. Production optics for10 Gigabit don't exist yet." "I believe that there are many systems companies that are waiting to see a certain level of maturity in the optics available from the optics component vendors before they complete their design efforts." "My best guess would be that 10 Gigabit we will see either very large numbers of 10/100 ports, or a significant number of gigabit ports which will be aggregated into 10 Gigabit pipes. I would expect that these ten Gigabit pipes would act as either cross connects or uplinks for each switch."

Thatcher is convinced that ten gigabit Ethernet will become the dominant form of long haul network transport and that it will eventually kill SONET. The question seems to be how soon. In the May 3 issue of Light Reading we found a ten part Report by Scott Clavenna on OC-768 40 gigabit SONET which is under development. Feeding the installed SONET base will be a viable proposition depending on what analyst you listen to for from anywhere from another year to between 3 and 4 years.

Thatcher: "In order to handle the consequences of what we're doing at the edge of network, I believe that we will see expansion of mesh topologies that inherently have much more bandwidth capacity. What may happen is that over the wide area and metropolitan area we will end up with architectures that are very stupid. All they will be doing is providing point-to-point high bandwidth connections. We will have situations not unlike the on and off ramps of super highways where there are a limited number of points that you can get on and off. If you want to go from where you are to where I am you find the nearest access point to the big pipe and boom, in one hop, you get to the big pipe egress point nearest me. If there is a break in the pipe, you want the intelligent optical network to fix the break in a way that is virtually transparent to you."

Frazier: "The grandiose picture that I draw is that if you have enough bandwidth coming into people's homes, and small businesses, and you can supply all their needed data telecommunications services to them in the form of IP packets over Ethernet then we get to the situation where Ethernet and IP become like a utility.

If you can supply enough bandwidth, at cheap enough prices, people will become very creative in their uses for that bandwidth. Such an achievement would enable new uses for data and data transmission. With data as a utility that is accessible as water, or electricity or gas or sewer, you will get uses for that data that you cannot currently envision." "The point is that with enough bandwidth coming into and out of the home, we won't care what the technology of delivery is. In such a case, I believe that we will see people begin to do a whole lot of things that we can hardly envision at this point. Ethernet in the First mile holds the potential to turn data use into something we take for granted like a utility. I predict that we will find that this will enable many new data network applications."

The Future of Telecom As Customer Owned Assets -- Bandwidth Derivatives Trading

p. 23

A short article on the continued growing interest in trading bandwidth as a commodity.

ENUM Pushes Convergence By Facilitating VoIP Access To Global PSTN Numbers

Rutkowski's Opposition Deflects IETF- ITU Plans Appeal To Free Market Is Delaying Rollout Of Services and Impacts Viability Of Business Use Of Voip Technologies

Mail List Debate Shows Significance Not Well Understood,

pp. 24 - 43

ENUM is a combination of address book and switchboard by which users of internet VOIP services will be able to reach any ENUM-registered telephone number and its attached telephone in the world.

An Internet Architecture Board IETF joint effort had worked out an agreement with the ITU to use a new domain e164.arpa as an address for connecting Internet telephony services to PSTN number and vice versa. The plan was for each nation state to decide how ENUM would be handled within its boarders while the ITU and RIPE would serve as delegators of e164.arpa country codes to each nationally chosen ENUM authority. Testing of the joint IETF ITU ENUM standard was to be underway by the spring of 2001. This hasn't happened.

Potential ENUM implementers are squabbling over the spoils, Network Solutions and Telcordia teamed together in October to launch proprietary ENUM services. Lead by Tony Rutkowski they have run a surprisingly successful campaign claiming to represent the unregulated Internet against the world's tired old telephone monopolies and regulators.

A discussion on the very astute Cyber telecom list in late April showed how surprisingly effective the Rutkowski forces had been in proclaiming to uphold the values of the Internet while blocking the implementation of an important IETF standard. The discussion on the Cybertelecom list effectively ended in a draw between the two sides where Rutkowski's forces had been successful in taking advantage of the new Bush administration's disinterest in telecom to checkmate further involvement by either Karen Rose at NTIA and DoC or by the FCC. As the discussion that follows shows - for the time being the IETF ITU planned implementation of ENUM is dead having been successfully ridiculed by Network Solutions and Rutkowski. Their determination to wrest it from the hands of the IETF and Neustar will likely mean that both they and Neustar will lose to the outcome of a pending struggle between the old line carriers, ATT and MCI on the one hand and the ILECs led by Verizon on the other hand.

The war of words taken from the Cybertelecom mail list and edited into the narrative that forms the majority of the article below end with a whimper. It does so in part because each side's debate is primarily ideological. Standards versus the free market. The IETF, ITU, Neustar side argues logically and persuasively from the point of view of standards, recognition of the fait-accompli that if one is trying to get the cooperation of the authorities which regulate the PSTN, it is irrational to try to win that cooperation by telling those with legal responsibility for national e.164 numbering plans to take a flying leap.

While folk like Patrik Faltstrom and Richard Shockey clearly have a more rational view of how to build the largest and most stable global market for ENUM, we contend that they have not carried the day, indeed may not carry it because they have not articulated in sufficient detail what can be don with ENUM they way that they wish to build it as opposed to the likely outcome of the Verisign approach.

We have read the April 24 (Vol 19, No 4) issue of Release 1.0 where Kevin Werbach does an excellent job of writing about ENUM in the context of convergence. We are given there profiles of four companies (Keen.com, Soundbyte, Pagoo and Acallto) with convergence dependent business models. The question becomes does ENUM enable convergence? People have talked about it as a tool that will enable the Internet to swallow the PSTN. The value of what Werbach calls the "Real Convergence Story" is that it shows us the necessity of thinking very carefully about our use of terminology.

ENUM is only an Enabler. ENUM does not control any kind of PSTN to Internet communication or vice-versa. It merely is an enabler that will work by taking PSTN phones numbers and putting them in a database so that the owners of the phone numbers can become users of these new kinds of services with the click of a mouse button if the SIP protocol has been added to the operating system of their PC. ENUM makes it easier for people to become VOIP users.

The ITU and national telephone numbering administrations are supposed to look favorably on a system for addressing new services for their customers that will be run by multiple companies operating under multiple rule sets designed to create a fragmented market. The system being advocated by Verisign is one where the certainty of global addressability for everyone choosing to use ENUM services has been handed off to competing companies whose motivation is not the coordination of a global service but rather the rapid registration of names to increase each ones market share. Such services would be saying to first time users of VOIP that they should trust their global communications to a fragmented system that could deliver its benefits by only trial and error and do so with out there being anyone who would take any responsibility for the protection of the identity of the person using the global telecommunications system. Migration of PSTN users into such an uncertain jungle would take years.

The phone number may be thought of as the closest thing to a universal personal identifier these days. What can be at stake with ENUM? Think of a world where government issued phone numbers to individuals and not phone companies. This is another way of thinking about ENUM. If a system were not in place that allowed individuals to assert and protect their control of their own numbers, their ability to communicate effectively would be seriously compromised. Under such a world it is highly unlikely that most people would trust the internetworking of their global personal identifiers to an unregulated free market based system.

The Carriers and ILECs have come to understand this. We may safely assume that THEY and neither Rutkowski on behalf of Verisign nor Shockey on behalf of Neustar will be able to dictate the outcome. As Larry Lessig has written: Code is law. How ENUM is implemented will determine who does or does not derive vast economic benefit.

Where ICANN Would Like To Push The Internet,

p. 43

An Internet controlled by the old style telecommunication and media giants with content and delivery increasingly unified. We reprint a part of a Gary Chapman column in which Gary writes eloquently about growing concerns of the state of the net but does not tie the motivation directly to ICANN.

End Notes:

Dave Hughes Blasts Alaskan Telephone Association Before FCC -- A Look At FCC Resources For Small ISPs -- A Critique of Tauzin Dingell,

pp. 44 - 47

We publish Dave Hughes April 18, 2001 FCC filing that delivers a well deserved grilling of the Alaska Telephone Associations position on the e-rate.

We also publish an exchange between Sean Donelan and Robert Cannon of the FCC. The exchange high lights the issue of a small ISP's problems with AOL Time Warner and the difficulties it faced in taking its grievances to the FCC. Very useful pointers for ISPs with problems in the future are provided.

Finally we take a quick look at the Tauzin Dingell bill HR 1542 "Internet Freedom and Broadband Deployment Act". Jeff Pulver found it to be " remarkably disingenuous" before making an unwarranted assumption that it makes VOIP illegal. We publish a critique of the bill by Fred Goldstein from the Cyber telecom list and note that the Cannon Conyers bill HR1697 and HR 1698 does deserve support.