We are finding more evidence that thephone companies are caught in an un-dertow from which they cannot escape -- unless the FCC and, or Congress doessomething truly idiotic like grant them amonopoly on fiber to the home. “Grantthem complete control over the glassand then they will invest” will run thetired argument. The first problem is thatthey have pledged this before and donenothing. The second problem is that ifthey were given yet another opportunitythere is and will be no enforcement forany of the pledges they make.

To borrow the metaphor from the 19thcentury, the result of granting them afiber monopoly would be to hamstringthe entire American economy into re-liance on “canals” in order to scare offthis new and chaotic world called “rail-roads.” While other countries are build-ing “railroads” - that is broadband - forus not to do so would irreparably handi-cap what is becoming one of the mostbasic infrastructures of a modern econo-my. We are already behind. The Orga-nization for Economic Cooperation andDevelopment (OECD) ranks UnitedStates only 17th in utilization of com-munication services. [Quoted in Feb 1P u l v e r.com letter to FCC ChairmanPowell.]

But fiber is fast you say and speed ofconnection is the issue – not control overaccess. What is wrong with such a sce-nario? Well consider the ZAP mail ex-perience as written by Clay Shirky on

January 7, 2003. http://shirky.com/writ-ings/zapmail.html What is wrong hereis not speed of connection but rathercontrol over the technology. The phonecompanies cannot see the world in anyterms other than those of control.

A Matter of Mindset --ZapMail and the Telcos

Consider what happened to the hot newcompany Federal Express in the early1980s. There was this new fangled de-vice called a fax machine that scanned adocument and sent the resulting digitalbit map over a phone line. The time wasjust before the split up of ATT and theexplosion of customer premises phoneequipment. Fed Ex totally missed whatwas happening. Thinking that its com-petitors were the other over-night deliv-ery companies, it spent 200 million dol-lars in an attempt to one-up them bybuying expensive new fangled fax ma-chines and building a dedicated phonenetwork to run them on.

As Shirkey writes in his essay, theyfailed to see that the breakup of ATTandthe consequent opening of the networkwould allow their customers to buy theirown fax machines and by being able touse the PSTN, become their competi-tors. They underwent a huge build outfor a business that wasn’t there. Withthe network opened up, Fed Ex’s cus-tomer bought thousands and then tens ofthousand and then hundreds of thou-

sands and eventually millions of fax ma-chines. Rather than rely on Fed Ex forthe faxing service, Fed Ex’s customersbought their own fax machines and didit for themselves. Today rather than relyon centrally controlled circuit switchedtechnology, increasingly large numbersof phone company customers are takingtelecommunications into their ownhands.

It is a simple matter of economics. Thecost of communication via IP is but afraction of the cost of doing it the phonecompany way. In our January-Februaryissue we saw how the large corporateenterprises are beginning to pull theirvoice service from the PSTN. Thisissue examines why Ipv6 is unlikely toever be significantly deployed in back-bone of the Internet. It also will showhow IPv6 deployed at the edge of thenetwork, in the hands of the end usercustomers of the phone companies,could do a great deal to redress the on-going consolidation of power into thehands of the central control minded tel-

On the Inside Tools forE d g e - b a s e d

Te l e c o m

Contents p. 2

Volume XI, No.12, March 2003ISSN 1071 - 6327

Building Tools for Edge Based ControlUnderstanding Edge IPv6 versus Backbone IPv6 Vo I Pand Vonage - When Customers Become CompetitorsOpen Spectrum Versus the Spectrum-as-Property Wo r l d v i e w

2

ContentsBuilding Tools for Edge Based Control -- Understanding Edge IPv6 versus Backbone IPv6 Vo I P and Vonage - When Customers Become Competitors - Open Spectrum Versus the Spectrum-as-P ro p e rty Wo r l d v i e w pp. 1 –4

IPv6 Going No Where - Political Push Fails to Propel Elegant Solution Lacking Market Pull - Former Drivers of Address Space, Device Addressing and Wireless Seen As No Longer Critical -While Very Important at Edge, v6 to See only Niche Backbone Deployment pp. 5 - 13

Is IPv6 Necessary? - One Year Later p. 10

IPv6 at the Edges -- IPv6 Seen Not as a Backbone orTr a n s p o rt Solution But Rather as User-Applied Edge-Based Overlay Supporting End-to-end A p p l i c a t i o n s pp. 11 -16

Two Internet Futures - With Edge IPv6 and Without Edge IPv6 p. 17

Customer Owned Networks --ZapMail and the Telecommunications Industryby Clay Shirky pp. 18 – 21

Discussion of Clay Shirky's ZapMail EssayUnlicensed, User Financed, Edge Based Connectivity Technology -- Locustworld Meshbox in Context of Building Edge Based Wireless Transport pp. 22 – 24

Open Spectrum - Property Rights World View Dies HardExploring the Problems with the Farber-Faulhaber Have-Your-Cake-and-Eat-it-Too Spectrum Arguments pp. 25 - 30

ICANN and the Failure of "Self Regulation"How the National Science Board was Overruled by the Clique that Became ICANN - Part One pp. 31 - 33

Governance by Lawrence LessigLessig Demonstrates How the Would Be "Self Regulators" Took Control - Part Two of How ICANN Came to Be

pp. 34 - 39

Interview, Discussion, and Article Highlights pp. 40 - 47

Executive Summary pp. 48 -51

cos and cable companies.

We have been learning a lot more aboutthe technology of VoIP. In our Aprilissue we shall return to VoIP and shallshow how new developments are alreadybeginning to lower the artificially highcosts of international phone tariffs. Weanticipate writing about the technologyand mechanics of the use of VoIP on aglobal basis as a substitute for expensivecircuit switching. The cost spread be-tween international circuit switched callsand VoIPcalls which can be routed fromone part of the PSTN to another is nowso huge that it has spawned a global greymarket. In part because so many peopleare so busy making money from it, this

grey market has never been discussed indetail in the press. Beginning with ourApril issue we shall do so.

The stark fact is that the blades of theVo I P scissors are closing in on thetelco’s cash flow. On the one hand oneblade is the result of large corporationswithdrawing voice traffic from thePSTN and running it over their corpo-rate IP networks. On the other hand theother blade is derived from internationalVo I P wholesaling by companies likeITXC and activities by thousands ofphone card middle-men hammeringlong distance rates ever downward. Theability of the phone companies to chargemore for a minute of voice traffic than

they could for a minute of data traffic israpidly diminishing. Recently the differ-ence has been as high as seven to one.That is if a telco could make a penny fora minute of data transfer, it could makeseven cents for each minute of voicetransmission.

For the most part the seven-cent differ-ential is no longer there. Bits are bits.One cannot really distinguish voice fromdata bits. That any price difference ex-ists at all is increasingly a regulatory ar-tifact. In two or three years market andtechnology pressures will have driventhe differential to zero. When this pointis reached, the telcos could find their rev-enues slashed by two thirds. They will

The COOK Report on Internet March 2003

3

than have all the relevance of Zap Mail.

Canadians AbandonFaith in Facilities BasedRegulation

Meanwhile the regulators cannot keeppace. Testifying before the Senate Com-merce Committee on January 14, Chair-man Powell said “The Commission hasbefore it a number of major proceedingsthat will attempt to improve and advancethe goals of the 1996 Act. With the bene-fit of hindsight, we will be able to assessthe last seven years and consider how wemight improve the regulatory environ-ment to more aggressively promote facil-ities-based competition, to promotemajor investment in advanced communi-cation infrastructure, and to reduce regu-lation—all hallmarks of the Act.” (p. 6)

As we showed last fall in our asset-basedtelecom issue (Vol. 11 Nos. 8 – 10), theCanadians have essentially given up ontrying to make facilities based competi -tion work. It is too bad that the FCC feelstrapped in the requirements of the 96 Act.As Powell describes it, the FCC is in dan-ger of striking out having taken twoswings and failed on both occasions. TheCanadians, realizing the impossibility ofthe task, have gone onto better things.

Powell has shown some interestingchanges during the past year. Amongthem is a journey from saying that hedidn’t know what the public interest wasto the following remarkable statement. “.. . we will be guided exclusively by thepublic interest, and resist the pressure toview our exercise as awarding benefitsand burdens to corporate interest.” (Pagei). Of course the proof will be not inwords but rather in actions.

On page four of his text he said: “In ad-dition, broadband connections have alsoput pressure on wireline networks asmany consumers that migrate to broad-band for their Internet services havedropped their second telephone lines(which were used for dial-up Internetservices). Moreover, 2002 saw the intro-duction of reliable Internet telephonyservices through a broadband connec-tion. Companies such as Vonage are pro-

viding consumers with a direct substituteto their traditional wireline phones.”“These various sources of competitionhave contributed to the first declines intotal access lines for the four majorILECs since 1933 (the only previous yearwhere access lines declined).”

Vonage and Cisco

With his mention of Vonage, we can cer-tainly see that Powell has more clue thanhe possessed a year ago. We are about tosign up for this service that for the firsttime takes a Cisco product (the AT186)VoIP gateway and treats it as a consumerproduct. The gateway plugs into the RJ-11 jack at the back of the phone and Eth-ernet into the cable modem in the back ofthe gateway. The result is unlimited longdistance in the fifty states for $40 amonth – plus very attractive rates to therest of the world. It is important to notethat Cisco, as a device selling companyand not a phone company, is well posi-tioned to profit from the VoIP price scis-sors.

Furthermore those who have read theNew York Times January 23 coverage(h t t p : / / w w w. n y t i m e s . c o m / 2 0 0 3 / 0 1 / 2 3 / t ec h n o l o g y / c i r c u i t s / 2 3 s h e r. h t m l) of ourwork with Tsering Gyaltsen Sherpa willsee another interesting aspect of whatcould be a new found Cisco view of theworld. Cisco donated Aironet 350 radiosto the Everest base camp project at DaveHughes’ urging. These are radios thatCisco markets to connect LANs inside ofbuilding in large corporations. Prior tothis it seems never to have occurred toCisco marketing people that these radioscan be used to bridge a LAN to a VSATat 5500 meters over a distance of twokilometers. The Times wrote: “Mr.Forster eagerly donated three Wi-Fi ra-dios on behalf of his company. Such ra-dios enable the creation of wireless net-works that can relay data within a coupleof hundred feet or as far as several milesas the crow flies, much the way thatlocal-area networks, or LAN's, work inoffices. "What I like about this project isthat it demonstrates that the technologydeveloped for a LAN in a building can beapplicable beyond that," Mr. Forster said."This may be as far outside the buildingas you can get."

For the first time Cisco is beginning tounderstand that these radios can be usednot just on a corporate campus but rathercan be used to replace the local loop incommunity based applications be itNepal or in connecting community net-works in Wales where Forster, at Hughes'urging, has also involved Cisco with pos-itive results. Of course, if they work inNepal, and in Wales, they will work inthe US. Unless acting in ignorance andon behalf of the telcos, our political andregulatory system forbids it.

The bottom line of all these events signi -fies only one thing. The local telephonecompany’s standard business model isdead. Rendered extinct by users takingcontrol of inexpensive technology andusing it for their own purposes. BecauseCisco, unlike Nortel and Lucent, alwayshad its major business outside that of thecarriers and the ILECs, it is in the wire-less and VoIPareas much better equippedto deal with the world in the aftermath ofthe death of the carriers than its moretelco-oriented sister companies.

We signed up with Comcast Cable Inter-net earlier in January in order to installVonage. On Monday January 27 we or-dered the package. The question of se-lecting the Vonage phone number wasnot immediately clear although withhindsight it seems obvious. The servicegives a separate phone line with its ownphone number. We still have our 609882-2572 number. Dial our new Vonagenumber 703 738-6031 and you will alsoring our desktop phone. Moreover if youare in the Washington DC suburbs and703 is a local call, dialing 703 738 6031gets you through to us for a local call re-gardless of whether you are a Vonagec u s t o m e r. The Vonage web pageshttp://vonage.com did not have an 800number listed for dial in. Frustrated. Wesent email asking to talk to a real livehuman. Not five minutes later our phonerang and a very helpful resident of Con-necticut who worked from his home aswell answered our questions. We com-pleted the order chose the Vonage phonenumber and were billed $40 for the firstmonth service, $30 for account activationand $10 shipping for the Cisco ATA 186gateway.

COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA

4

The package arrived the next day. Theonly hassle was buying a two line phone(price range 30 to 60 dollars) and theright RJ11/14 connector. Installationwas a breeze. Total plug and pay. Wellalmost. We signed up for Free WorldDial Up and that does not work withVonage. Why? Because the gateway isshipped pass word protected and to workwith FWD a proxy IP number must beinserted.

The Cisco gateway retailing for $200and available wholesale at $135 and“free” from Vonage is very compact.Roughly 6 inches by 6 inches and lessthan 2 inches high. Plug in the powercord. Plug in the RJ11 from the phoneand connect an Ethernet cable from thegateway to our 8 port fast Ethernetswitch ($45). With line 2 on the phoneset as the default (the VoIP line pluggedinto the gateway) pick up the phone anddial. The gateway has a large red lighton the top that glows red when working.

The experience actually was totally plug

and play! It was not even necessary toopen a network panel and configure anIP number for the gateway. Our AppleAirport Base Station did that transparent-ly acting as firewall and router. Theyquality is excellent. Our first interna-tional call was to Arcady Khotin in StPetersburg, Russia. The cost was sevencents a minute. The same rate that wewere paying ATT for domestic long dis-tance.

The Center is Dead

The center is dead. Forward movementis at the edges. The major focal point forthis issue is IP v6. Farooq Hussainshows why its chance for significant de-ployment in backbones at the core of theInternet is effectively zero. However ina discussion with David Reed, BobFrankston, Francois Menard and Farooqwe are introduced to the concept of V6 atthe edge of the network. We begin to un-derstand how V6, in the hands of endusers at the edges of the network, couldredress the shift toward the center thathas taken place in the balance of control

within the Internet. Indeed we havebegun a fairly in depth exploration. It isnot yet really clear what Microsoft willoffer in order to make edge based IP v6applications plug and play. Standardswould help enormously. Five to tenyears ago the IETF would have been theplace to turn. Today it might be theIEEE.

Or it might even be the Consumer Elec-tronics Association. At some point, wehope to offer input from Vi rg i n i aWilliams who is active at CEAin leadingan effort to enable whole families of de-vices plugged in at the edges of IP net-works to find each other. In a conversa-tion with her on January 30 we learnedthat there are several consortia of com-panies within the consumer electronicsfield that are exploring a range of issuesthat could be described as loosely relatedto Edge Based v6. We hope to describethese efforts in more detail in a futureissue.

Editorial Calendar

In the next issue we shall return to Voice over IP.In the one after that we likely shall do a reprise ofasset based telecom which is now going global inmajor ways.

5

Highlights

Editor’s Note: Farooq Hussain was thePrincipal Investigator for the SprintN A P and moved shortly after theNSFNET transition from Sprint to MCIjoining the team directed by Vint Cerf.He left MCI just prior to the completionof the merger with WorldCom havingworked on both the merger plan with BTand subsequently WorldCom for the In-ternet components of MCI. He was withAGIS for a little over a year helping toestablish a business relationship withTelia of Sweden who subsequentlybought AGIS out of bankruptcy. Cur-rently, he is a partner in a research andconsulting firm Network Conceptionstogether with Phil Jacobson [also an ex-MCIer]. We interviewed Farooq on Jan-uary 3, 2003.

Is IPv6 a DeployableProtocol?

Hussain: IPv6 and the question its de-ployment is wrapped up in a series ofquite complicated tensions which aredifficult to articulate. My interest is infocusing on the policy issues that sur-round it as well as the lack of any rea-sonable way to determine what the com-mercial value of deploying it would be.There are two camps. One says IPv6 isnot needed and won’t happen and thosewho say it is absolutely necessary andwill happen. These diametrically op-posed positions all stem from a veryfundamental issue of where we are withprotocol development.

Two years ago a major international car-rier whose networks were certainly ap-plicable to IP v6 commissioned me todevelop an IPv6 strategy for them.(This carrier has operations in Asia, Eu-

rope and North America and is financial-ly stable.) I hadn’t paid a lot of attentionto what had been going on with IPv6 be-fore mid 2000 or so. Like everyone elseI had been reading all the announce-ments that it was “about to happen” andmy first inkling was that as long as it wasabout to happen, perhaps this clientshould be doing something about it.

At the time there were three or four largeUS Operators, most notably WorldComand Sprint who were saying that theyhad v6 networks operating.

COOK Report: In the sense of test net-works or were they really production?

Hussain: I believe the old vBNS had v6going. You had in the engineering com-munity a lot of tension between thosewho were strong proponents of v6.There was some middle ground amongthose who were not really bothered oneway of the other while on the other sidethere were and still are some very verystrong critics of v6.

There has been, from the very beginning,a considerable amount of tension withinthe IETF about the need for an approachto IPv6. Lying at the very foundation ofan understanding of where v6 is going isthe necessity of understanding the ra-tionale for its creation back in 1992 - 93.Everyone said then that we were goingto run out of address space. This con-cern about address space continues up totoday to be stated as the key rationale forIPv6.

The reality is that the problems with theInternet protocol that v6 was designed tosolve have been managed during thecourse of the intervening decade bothwithout v6 being available and without ithaving become a convincing alternative

to the existing v4. Some critics nowwould say that part of the problem is thatthe whole goal of expanded addressspace is just propping up the establishedconcept that every device reachablefrom the Internet needs at least one per-manent layer-three address.

The Presumed AddressSpace Shortage

Ten years ago this was actually not suchan unsound approach. We then had thisidea that the car would have its IP ad-dress and that within the car maybe theair conditioning system and carburetoralso needed their own IP addresses. Justas every house has a phone number,everything was to have its own IP ad-dress. But things have turned out ratherdifferently. We are much more sensitiveto devices and uses being session orient-ed. And having, as a result, temporaryaddresses.

Now we are looking at problems of theInternet in going forward a decade laterand it will not necessarily be appropriateto say that what has happened over thepast 10 years to the way that v6 has de-veloped actually applies very well to thecurrent situation.

COOK Report: In terms of current op-erational economic and technology con-cerns?

Hussain; Exactly! On all levels! Butthe difference of opinion in the engi-neering community is really substantial.Seen in this light we have had a parallelpath of the pursuit of the development ofv6 while, at the same time, IP Sec,MPLS, NAT all of these things, let alonethe management of address space, havehappened and, in their respective ways,

IPv6 Going No Where - Political Push Fails toPropel Elegant Solution Lacking Market PullFormer Drivers of Address Space, Device A d d r e s s i n gand Wireless Seen As No Longer CriticalWhile Very Important at Edge, v6 to See only Niche Backbone Deployment

COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA

6

have extended the viability of v4.

COOK Report: And there is now a lot ofinfrastructure in place that depends onwhat people thought might be only tem-porary patches. These patches are in factnow turning into very permanent lookingfixtures.

Hussain: Yes. As time passes, it be-comes less and less appropriate to call itonly a patch. I think that if you sudden-ly started telling people that NAT ad-dressing is only a band-aid, they’d lookat you as though you were more than abit loony. It is here and working fine.When you get into these discussions, youhave arguments that are about issues oftechnical and architectural elegance.People will look you in the eye and saybut v6 was designed to have security asan integral component. It has auto con-figuration as part of its design. We knowall this but if we look around we areforced to acknowledge that it still isn’there. It has a lot of nice “features’ - yetpeople still are not using it.

COOK Report: Shades of OSI! It is theoutlook that says I will promise youeverything if only you are patient.

Hussain: It definitely is afflicted withbits of OSI. But the road to v6 startedout in a fever pitch rather more like themarch to Y2000 fixes because everyonewas propelled forward by the idea thatthe exhaustion of address space wouldkill the new-born Internet. Also whatmay prove to be the most damaging thingfor IPv6 is that governments have man-dated its use. One might ask why onearth they would do this? Why wouldthere be official political battles, at thenational level on behalf of a communica-tions protocol?

Institutional Proponentsof v6

The main source of institutional supportfor IPv6 now in the US is to be found al-most exclusively within the Departmentof Defense. No one else really battles forit. But even then it is really difficult tosay exactly how strongly DoD is reallypushing it. Someone has made a deci-

sion to support it and, whatever the rea-sons for doing so, are not really com-pletely clear to me. The other parts of theUS government don’t seem to care.

In Europe it is very strange to see that theEuropean Commission is hugely in sup-port of v6. They have quite a few initia-tives, including a couple of major ones,on-going to push forward the protocol.Meanwhile Japan has long been in favorof v6 and indeed has become the onegovernment to actually mandate v6. Youhave then a significant portion of theOECD countries in terms of their respec-tive economic power who are in favor ofv6. But looking at the over all situation,you must say that the US is not quitethere. That Japan, from the governmentperspective, is totally pushing it. Europeis trying to push it and, in fact, there is aninternational alliance between the Euro-pean Commission and Japan to endorseand promote IPv6.

But looking at all this official supportyou need to ask what is going on here? Isit not good enough to get adopted on theface of things? It is rather unusual tolook at a protocol and proclaim thatsomehow it is the key to some economicpower. Or that it will lead to some terrif-ic economic advantage. It seems to methat this outlook is one that fights the lastbattle. It says that the US gained greatadvantage from IP v4 so let’s try to gaincomparable advantage from being thefirst with a replacement for v4.

I am not at all sure that this makes muchsense anymore because the rational forv6 is about controlling and managing ad-dress space. Where you find the heaviestendorsement of v6 is where the routingregistries have the most severe policies.Japan certainly falls into this category.APNIC pushes v6, but within APNIC,Japan pushes especially hard.

In the initial allocation of v4 addressspace, the claim was and remains that theUnited States allocated address space insuch a way that certain countries wereleft very short changed.

COOK Report: If you had a Class A ad-dress block and many universities didand still do, you had more address space

than all of China.

Hussain: Quite true. Take therefore theJapanese position that Japan is very tighton address space and that it is requiredtherefore to manage it very carefully.Consequently it is going to mandate theuse of v6 because doing so frees Japanfrom any constraints imposed by the ar-bitrary nature of the way in which the ini-tial allocations were made.

The Position of Japan

In this context the most important paperthat I would direct your readers to is onethat they really should read before get-ting absorbed into the detail and finer is-sues of why v6 is in my view unlikely tobe anything more than a niche protocol.This paper was published by Glocom inJanuary of 2002. Its title is “Is IPv6 Nec-essary?” It is by Nobuo Ikeda and Ha-jime Yamada. Seeh t t p : / / w w w. g l o c o m . o rg / t e c h _ r e v i e w s / t e ch_bulle/20020227_s2/ The paper is wellput together with a very balanced argu-ment. But note also that it is from Japan!

The authors estimate that we are unlikelyto run out of v4 address space for anoth-er 15 years – if ever. I haven’t seen thispaper really challenged. When I read thepaper, I wondered what would be the ECreaction? Would the EC just quietly de-fuse its support? There has been an enor-mous push back from European ISPswho fear that they might be mandated todeploy v6 just as ISPs in Japan were. InEurope there is push back against the ECdirective as well as all the hype that youhear for it. What I do see is that, in Eu-rope, the conclusions of the paper arebeing wished away.

Since the Ikeda -Yamada paper is basi-cally a research paper, the proponents ofv6 breathe easier knowing that it won’tfall into the hands of the trade press thatgoes on cobbling out simplistic argu-ments that we better hurry before addressspace is gone and the huge numbers ofwireless users all of whom will have de-vice dependent IP addresses arrive. Allthese assertions go unchallenged exceptwithin that core community that had seri-ous issues with v6 from the very begin-ning.

The COOK Report on Internet March 2003

7

COOK Report: The people with the is-sues are those who have v4 infrastructurein place, are running happily and do notwant to have to make the huge invest-ment in changing?

Hussain: Yes. But furthermore the hugeinvestment in changing would require arationale propelling the change. Whywould we be making a huge financialcommitment? What would we anticipateour return to be on such an investment?

If you have a large network and are re-quired to implement this protocol, youcan derive an operational or internal ben-efit. Or it can come because there is mar-ket pull. It is something that customerswant. Now we have been told that cus-tomers will want v6. But the window ofwhen customers will really want it hasbeen moving outward now by 2 to 3years every six to eight months.

Window for AllegedMarket Pull KeepsReceding

When I first started looking at this in theyear 2000, the period 2001 –2003 wasgoing to be the big and explosive periodof IPv6 adoption. Two years later we arelooking at a period of somewhere be-tween three and five years before there isany indication of a recognizable marketpull in the wireless arena. Projected pullthat is 3 to five years distant is somethingthat is too uncertain to be a reason for usto commit to capital expenditure now. Inshort I think it quite safe to assert thatcurrently, there is no reason to deploy v6because of market pull.

There are ways to implement v6 as tun-neled within v4 within a backbone net-work. You might consider doing this as ameans of gaining experience with it as aprotocol. Most players out there who saythey have v6 are implementing it in thissort of marginalized way. When youlook at what operational benefits are tobe gained by turning a backbone networkat the Internet core into an IPv6 network,there are really precious few. To turn abackbone network into a v6 network,there are actually quite a few levels of

complexity to undergo. To arrive at v6you will need to do serious levels of pro-tocol translation at the edges because ob-viously all but a negligible fraction ofyour traffic will be originating and termi-nating as v4. From an operational stand-point, as a large network, saying v6 doesthis and that better than v4 for me makesno sense because no such a network livesin isolation from the Internet. You haveto be dealing with v4 anyway and whatyou end up with therefore is in effect adual direction that is now being pursued.

So where are we now? I would say thatv6 is a pretty solid protocol. There is a lotbeing done to address transition. Most ofthe key core router manufacturers –Cisco, Juniper – and a couple of otherssuch as Hitachi have announced releasesfor v6. They are basically offering theirrouters with dual stacks. Networks thatdeploy v6 will be doing so with dualstacks. This means that you will have v4,v6, MPLS, and must have a dual stackDNS - in short you will have a lot morecomplexity to deal with. Heading in thisdirection does not mean that you havechosen a path to operational efficiencyand cost savings in the core of the net-work. But in tough economic times thisis the direction in which everyone musthead.

COOK Report: Well suppose a universi-ty wanted to operate v6 only on its cam-pus? But even doing it just on its owncampus would increase the cost of opera-tion?

Hussain: You have to ask just what it isthat they would gain from v6? Do theyneed to run v6 because they don’t knowhow to do NAT? Or because they won’thave enough address space? When I wasevaluating v6, I found a very ambivalentposition on the part of educational insti-tutions. The 6 Net that has six or sevenhundred institutions is hosting the net-works in general of small research de-partments. I really don’t think that thesedepartments are representative of thecampus network of the entire the univer-sity.

The bottom line is that we are having aproblem in finding a commercial ration-ale for deploying v6 solely on the justifi-

cation that it is more elegant than whatwe have as an alternative. The debate be-tween elegant v6 versus plain old v4 isbeginning to bear the marks of the dis-putes of MACs versus their PC brethren.There is precious little that v6 a decadeago was designed to do that cannot nowbe done in other ways. You can almostcertainly say that there are some thingscannot be done with v4 in ways that areas elegant as those to be afforded by v6.The problem was that v6 has simply notbeen there for other purposes because itswhole design rationale had been drivenby the warnings of v4 address space ex-haustion.

COOK Report: All the talk was of the 60MPH collision with the brick wall whichbecause of Cider and DHCP didn’t hap-pen.

Hussain: Don’t forget NAT. All of thishas become part of a fabric that is globalin scope. If you now try to envisage atransition to IPv6 set against this existinginstalled infrastructure of v4, I think theGlocom paper not sarcastically suggeststhat it will take centuries. If there wassome market pull, one might say there isa rationale for it to happen.

Isolated Rational –Wireless 3GPP

I think the rationale for IPv6 exists onlyin very small isolated cases. Let me lookat them by putting the small isolated casethat is the most contentious of the lotfirst. Wireless. The wireless environ-ment has really had a number of interest-ing twists and turns. V6 has taken adecade to declare that it has solved theaddress space problem by essentially giv-ing everyone infinite space. But in paral-lel our way of handling address space hasbecome so good that we no longer needthe solution that v6 has labored so long toachieve. It is very unclear that we havean address space exhaustion problem thatcannot be managed. Moreover we havemanaged it quite well so far.

The other issue is why does every deviceneed an IP address and the conclusion isthat it probably doesn’t. So put these twothings aside and look at what you have.You have networks that are carrying

COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA

8

IPv6, IP SEC, MPLS and Ipv4. My con-tention is that in the future IPv6 will bethe smallest niche component of this traf-fic.

COOK Report: But once upon a timewireless devices were going to needfixed addresses. Do we now have theequivalent of DHCP for wireless?

Hussain: I think the situation aboutwireless is fundamentally unclear andquite contentious. 3GPP, which is thethird generation mobile project, adoptedIPv6 as their protocol of choice in 1999.In doing so they probably gave v6 thestrongest endorsement that it has ever re-ceived. It claimed that each cell phonewould have its own IP address and thatthere would be billions of handsets. Therequirement for using IPv6 to handlesuch addressing issues seemed to make alot of sense. But there were a couple ofproblems.

COOK Report: For one until a cellphone becomes totally digital it doesn’tneed an IP address. Right?

Hussain: Correct. And furthermore theymay never get to that point because thereis something else going on with the wire-less operators in terms of their selectingv6 as a protocol. The mobile operatorsand certainly those outside the UnitedStates have been very pleased, and right-ly so for that matter, in terms of theirability to establish mobile roaming.When they approached third generationroaming requirements for data, it wastheir intention to have a third generationwireless network run as an IP network.But their idea was there would be the oldInternet and a new 3GPPInternet with itsown addressing and its own domains. Ifyou want to send traffic to it (3GPP) youwould have to connect to it and peer withit.

There was a moment in time during theheight of the bubble when for the blink ofan eye you might have said “my godthese people are trying to compete withand over take the global Internet with oneof their own construction!” They simplydidn’t seem to understand the most fun-damental points of what they were deal-ing with in terms of the Internet. On top

of all this, at some point the mobile op-erators decided to have a competition toestablish exchange points for mobile In-ternet operators that were also deliveringother kinds of Internet services.

If you were an Internet operator youcould have an exchange agreement.(Cable and Wireless, the Amsterdam AmSix, Sprint were among those involved.)A whole bunch of mobile operators gotinto this group that was interested in cre-ating exchange points. These mobile op-erators were trying to create an insulateddomain that was outside of the manage-ment of the routing registries. The ef-forts never really took hold. No bodycomplained about it but also nobodypointed out that it was really a veryflawed approach.

COOK Report: They were adding anoth-er layer of complexity.

Hussain: If mobile Internet had actuallystarted to take hold, I think they wouldhave seen a problem of huge dimensions.

COOK Report: Why?

Hussain: The exchanges we have noware just hanging on. Segmenting themarket further into exchanges for justmobile operators would not I think havemade much sense. I think there were fac-tors at work here beyond just ones of get-ting IP connectivity to your cell phonethat had slowed up and disrupted thingsin the mobile market place. This slowingand disruption was I think a fortunateside effect for those of us concernedabout the Internet’s strategic direction.

So now what we actually have to ask iswhether it will be 2007 when 3GPPstartsto happen and we are all going to have tobe ready with V6 because this is the kindof forecast date they are asking us to lookat right now.

COOK Report : But if we have softwareradios coming on line right now by thenwe shall have software defined cellphones.

Hussain: Exactly. At the beginning of2001 they were talking 2005 at the be-ginning of 2002 they were saying it was

going to be 2007.

COOK Report: By 2005 your cell phonewill sync to other cell phones in theneighborhood and likely be able to figureout what kind of address grid it is in. Inthis sense a geographical addressing sys-tem could become possible?

No Impact Before 2007But by Then WholeNature of Wireless WillBe Changed

Hussain: Exactly. Other than this beliefthat we are going to run out of addressspace, the only rationale for v6 is that weare some how going to have billions ofmobile users whose operators are goingto need v6. My analysis of this hasbrought me to the estimates of others thatclaim by 2007 mobile requirementscould make an impact. The problem isthat by 2007 there will likely be enoughother changes in the way mobile workssuch that no one else will want theseIPv6 related capabilities.

COOK Report: Because there will beother better and cheaper ways of doingit?

Hussain; And these are already showingup now. The compelling arguments forv6 are based on two things. Addressspace considerations and mobile devel-opments that might represent an uncon-trollable growth problem that would ex-acerbate the address space issue.

COOK Report: If Powell carries theopen spectrum reform forward, historymay show that it was this effort that ren-dered IPv6 unnecessary.

SONY Proclaims v6

Hussain; Precisely. However, here is afinal issue. About 18 months ago a highSONY executive declared that all futureSONY devices would be IPv6 address-able and warned that all service providershad better deploy v6 to be ready to takeadvantage of Sony’s roll out.

The problem is that even if SONY’sstrategy were to work, those v6 devices

The COOK Report on Internet March 2003

9

would have to talk v4 as well because allnetworks will never convert over night.In order for the device to operate howev-er the protocol must be transparent and itwill have to work as well on v4 as on v6.You are just adding another level of com-plexity with no real payoff since if it willoperate v4 there is no real need for it tooperate v6.

COOK Report: So even if the networkswere impressed enough to start investingand even if they had enough cash to do so(which they do not), the issue off addedoperation complexity would suggest thatthey don’t go forward with v6?

Hussain: Yes. In a couple of years youwill see all the major core routers withdual v6 and v4 stacks. You will be ableto serve customers by setting up v6 tun-nels inside of v4 for those customerswhich for some reason or other just haveto have v6. You could also use MPLS toset up a native v6 PVC. But the aggra-vation to do this is extensive and even ifit were cheap, the idea that your engi-neering team will be eager to rush outand embrace v6 just isn’t likely.

We may well be faced with quite an ironyif we are faced with the need to run twoversions of IP on the internet – v4 whichwill likely never go away and v6 forwhich there may be a few niche markets?IPv6 was intended to replace v4. It is un-likely that it will ever achieve this goal.But it certainly has its advocates and itsniche applications. Given the current di-rection in which we are going we will nothave a permanent address for every de-vice as envisioned in v6. There could besome circumstances that include the pos-sibility of a global 3GPPnetwork that in-sists on having fixed IP addresses forevery device dependency. So instead ofone IP protocol to be managed you will

now have two.

One has to ask whether the purportedbenefits – address space, security andauto-configuration are worth it. Do thesebenefits outweigh the aggravation ofhaving to manage two versions of IP inthe network? In other words IP Secwould be nice, but if the cost of getting iton a meaningful scale is a multi billiondollar global reconfiguration program,are there other less expensive ways of en-suring security? The answer is very like-ly yes.

Still No Market Pull

COOK Report: At one point a good v6stack in Windows was supposed to bringon the v6 revolution?

Hussain: The irony is that a lot of peo-ple have now come out with good v6stacks and the existence of these stacksisn’t doing anything for anyone.

COOK Report: Once upon a time youwere going to use SIP to be able to turnon and off your home air-conditioner.Now however the air-conditioner and abunch of other stuff sit behind your homefirewall and you don’t want it to be uni-versally addressable?

Hussain: That is about where we are.These applications are suggested asthings that are doable with IPv6 but theyare not market driven applications.There is no evidence that there are largenumbers of folk out there who want to dothis.

COOK Report: At one point people werecomplaining about firewalls holdingback end-to-end v6 capable architecturesbut if we didn‘t have firewalls out thereprotecting our cable modems from Klez

worms what would we do?

Hussain: Exactly! There are pluses andminuses on both sides but from the per-spective of a network operator the worstof all worlds is emerging in that theyknow that they will have to deal with v4for the vast majority of their customerand transit traffic but that there will alsobe certain circumstance in which theywill have to envision carrying v6.

They will have to manage both and theymay or may not make a decision in acouple of years time that their core net-work would be v6.

COOK Report: Is there any reason whythey might go to v6 in the core?

Hussain: Only because they routers arealready enabled and they could do itwithout having to spend any significantextra money. Doing it would increasecosts and add complexity and in the ab-sence of market pull it is unlikely thatthey would do it. Right now I don’t seea pull.

DREN did exists as a defense researchnetwork run originally by ATT. They lostit and Global Crossing had picked it up.But when Global Crossing went bank-rupt more than a year ago WorldComhaving had in the vBNS experience inplaying with v6 won he contract. But theonly market pull here is coming from theUS DoD that, as I said earlier, is the onlysubstantial advocate for v6 in the USA.DoD might indeed want vast numbers ofdevices with fixed v6 addresses. But theDoD's requirements are likely to have amilitary rationale and not one that willtranslate into creating a broad commer-cial market pull for the general imple-mentation of IPv6.

10

COOK Report : We asked Nobuo Ikeda,one of the authors of Is IPv6 Neces-sary?” h t t p : / / w w w. g l o c o m . o rg / t e c h _ r e-views/tech_bulle/20020227_s2/ to givehis opinion on the subject of his paper ayear later. He replied

Ikeda: I suggest you ask comments fromJun Murai, the global leader of IPv6. Heand I discussed this problem in a recentIETF meeting and agreed that the prob-lem is not the "shortage" of addressesbut rather applications to take advantageof v6.

COOK Report: May I send him yoursuggestion?

Ikeda: Yes, but I'm not very much inter-ested in v6 because we have reached aconclusion. Yet it's important to make itclear because there are still many peoplewho believe in v6 without knowing it.Our paper on IPv6 was downloaded by17000 people last year.

COOK Report: In other words v6 isdead?

Ikeda: I think it is still alive as an ideal.

As Larry Lessig emphasized in his book"The Future of Ideas", it was the E2E ar-chitecture that made possible the explo-sive innovations on the Internet. Buttoday it has drifted far away from thisprinciple. The Internet is complicated,opaque, and controlled by serviceproviders. Can we take it back to theE2E ideal? I'm not optimistic about that.Today, five years after RFC 2460 thatrecommended IPv6, the number of v6sites has increased to 1259 from 1046 inNovember 2001 - about 180 sites peryear among more than 40 million siteson the Internet. How many millenniumsdoes it take to replace v4? Seeh t t p : / / w w w . c s -i p v 6 . l a n c s . a c . u k / i p v 6 / 6 B o n e / W h o i s / i n dex.html#full

Some people argue that "ubiquitouscomputing" requires IPv6. However, it isnot v6 but Auto-ID that is prevailing asthe international standard for RFID. AndI don't think it would be so ubiquitous asthey imagine. Indeed NAT is ugly, but itprotects average users from direct at-tacks on their IP addresses. We evendon't receive e-mail by E2E. It's the real-ity of the Net whether you like it or not.

The cheapest way to recover the E2Ewould be to reallocate v4 addresses ofwhich only 3% are used. I recommendthat ICANN to have "address buyouts"to buy back idle addresses by reverseauctions and sell them through auctions.[Editor: Not trusting ICANN, we areglad that the chances of this happeningare quite remote. But as a general prin-ciple we see the point that Ikeda makes.]I proposed a similar mechanism foropening spectrum in my article "TheSpectrum as Commons".

h t t p : / / w w w. r i e t i . g o . j p / j p / p u b l i c a t i o n s / s ummary/02030001.html

Indeed the problem with IP addresses issimilar to that of the spectrum "short-age". In fact, spectrum is not runningshort; it is only monopolized by incum-bents who can't use it efficiently. Internetpeople are accusing the incumbents ofstifling the innovations made possible bynew radio technologies. I would make asimilar accusation for MIT, Apple Com-puter, Hewlett-Packard, and other organ-izations that each have more addressesthan all of China.

Is IPv6 Necessary? - One Year Later

11

COOK Report: Asking for comments,we sent the Interview with Farooq to BobFrankston (one of the developers of Visi-Calc), and to David P Reed, and FrancoisMenard whose names should be quite fa-miliar to our readers.

On January 20, 2003, Bob Frankstonreplied: It's very important to distinguishbetween V6 at the edges and V6 in thebackbone. The reason that V6 is not cur-rently available is that those who are theguardians of the net -- the backbone peo-ple are just worrying about their internalissues and there is no concept hereof ac-tually using the network.

Here is what I wrote last summer: EdgeProtocol (EPv6) rather than IPv6h t t p : / / w w w. s a t n . o rg / a r c h i v e / 2 0 0 2 _ 0 6 _ 3 0_archive.html#85208157

I recently (June 21st, 2002) spoke at theIPv6 (Internet Protocol version 6) sum-mit (h t t p : / / w w w. i p v 6 s u m m i t . c o m / i p v 6 -program.html). I was invited to speakabout the issues raised in my essay on theImportance of Encrypted IPv6. In thatessay I pointed out we need to assure thatevery system connected to the Internethas its own (IP) address so that it can bea full peer participant. Encryption is im-portant because the separation of the ap-plication layer (TCP) and the transportlayer (IP) has been weakened byproviders who are second-guessing thetraffic on the network.

Despite the urgency there are many whowonder if we'll ever be able to make thetransition from IPv4 to IPv6.

The answer is "no" because that is thewrong question. The idea of transitioningthe entire Internet to a new protocol rep-resents a failure to understand that the In-ternet has thrived because it is defined byits users rather than by a central authori-

ty. IPv6 has been designed as a protocolthat tries to meet the needs of the user(application) layer and the transport layerat the same time. While IPv6 does a rea-sonable job at meeting both requirementsthe deployment model is seriously flawedbecause it ignores the dynamics of the In-ternet as a marketplace driven by theneeds of each user. [snip]

IPv4 (or just "IP") represented the birthof the Internet by shifting the power todefine the network to the users at theedges.

The Internet has thrived because supplyis driven by demand. New applicationservices are supported by simply provid -ing more transport (or IP) capacity.Rather than wait for new capabilities tobe defined, users will create their own so -lutions. (When I say "users" I don't meanall users create applications. It only takesone motivated, creative individual withsome time on their hands to create an ap-plication that will be adopted by millionsof others. We just don't know which userthat will be.)

[snip] The solution is severing the de-pendency upon IPv6 as a way to meet theneeds of the transport layer. Instead weneed to focus on the requirements at theedge of the network.

Edge Protocol 6

I'm proposing a new protocol called EdgeProtocol 6 to give us the benefits of thelarger address space and simplicity. Itgives us the ability to make immediateuse of IPV6 technology at the edgesusing the Internet as-is.

We must not lose sight of what is reallyimportant, namely recovering the sim-plicity of the Internet by giving each endpoint a public presence. By implement-

ing security between end points not onlydo we have a chance of understandingwhat is happening, we can also chooseour own policies. Barriers between sys-tems (including firewalls) seem more fo-cused on fear than on allowing organiza-tions to create value.

There is no requirement that the edgeprotocols and the transport protocols bethe same. It should be consistent andconvenient to leverage common formats.Those of us at the edges have alreadypaid a high price in waiting on those of uswho are tweaking IPv6 for use within thebackbone for the Internet. This continuesto be a dysfunctional dependency. Wemust learn from the success of the Inter-net itself and treat the relationship be-tween the IPv6 and what I am callingEPv6 as similar to the separation of UDPfrom IP or IP from Ethernet packet for-mats.

Though this is really Edge Protocol ver-sion v1, I am calling it v6 for marketingreasons and it looks a lot like IPv6. Thebig difference is in the requirements. Wecan deploy EPv6 on the existing IPv4 In-ternet now. Not only does this avoid de-pendency upon unmotivated serviceproviders, it also allows us to ignorethose who are trying to build out an IPv6network since we shouldn't care whethertheir efficiencies come from adding ca-pacity or clever protocols. In fact, weshould discourage any cleverness infavor of just adding capacity.

Key EPv6Characteristics:

• Supports a larger address space. Ad-dresses can be composed using the exist-ing IPv4 addresses as a prefix so we canuse the existing infrastructure.

• Address resolution can use the existing

IPv6 at the Edges IPv6 Seen Not as a Backbone or Transport Solution ButRather as User-Applied Edge-Based Overlay SupportingEnd-to-end Applications Highlights

COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA

12

IPv4 DNS Entries (A records) as well asthe newer AAAA (IPv6) records. Thus Icould say"rmf19.myhouse.frankston.com" where"myhouse.frankston.com" is a V4 ad-dress part.

• Connections are assumed to be encrypt-ed in order to discourage favors fromthose who are fixated on "efficiency" andother meddlers.

• While there are elegant approaches tothe problems of NATs (those routers youbuy for your home), EPv6 implementa-tions can even fall back to TCP tunnels.The advantage of TCPis that it maintainsa connection through NATs but the dis-advantage is that it can impose arbitrarydelays and overhead. From a marketingstandpoint, however, it means we can useEPv6 without changing the NATs andthat can create a demand for better solu-tions. The risk is that the pain won't begreat enough to force a change but thatisn't all that bad.

• EPv6 is meant to enable new applica-tions. Transitioning existing services is asecondary priority though being able toaccess EPv6 web sites from older sys-tems is important but can be done at theapplication level.

The Internet has been seriously weak-ened by the need to share IP addressesamong a set of computers. We're tenyears overdue on remedying the situa-tion. The availability of EPv6 is a keypart the rebirth of the Internet. The P2P(Peer to Peer, including Instant Messag-ing and other collaboration tools) com-munity already represents a significantpent up demand that is ready to catalyzearound a commonly accepted way to pro-vide a large address space with directconnectivity between systems at the edgeof the network. The use of encryptionhelps assure that the connection is indeeddirect.

Transport providers who do want to takeadvantage of the IPv6 addresses to sim-plify routing will also benefit by having ademand for their services. The processwill start by building on IPv4 but theability of EPv6 will also make it easier tomeet the demand using IPv6. IPv6 with-

out such a demand isn't very interesting.

V6 in the Backbone andV6 at the Edge isEntirely Different

On January 20 Bob Frankston remindedus in response to the IPv6-in-the-back-bone focus of our interview with Farooq:

The purposes of V6 in the backbone andV6 at the edges do not have any relation-ship whatsoever. Period. No qualifica-tions. This has lead to the tragedy of themisperception of a commons. The back-bone has indeed accommodated itself toV4 since trying to address each atom onthe net individually is a very big problemand unnecessary. The IP "address" is likethe circuit ID in the phone network and itencodes a routing though not necessarilya precise one.

V6 in the backbone has become a feedingfrenzy for those who miss the PSTN andwant to bring back QoS (AKA discrimi-nation in favor of legacy traffic and tojustify maintaining scarcity) and MPLS(circuits are forever). There's also the badidea of providing mobility at the IP lay-ers. (Yes, saying this is in conflict withcomplaining about temporary IPaddress-es but that's a longer discussion).

There is indeed no market pressure formthe sheep at the edges so there is no wayISPs will make it a priority and waitingfor them is pointless. The reason we needaddresses at the edges is to give everyend point a first-class public presence onthe net with a modicum of stability.NATs, and VPNs and Firewalls have al-ready destroyed the Internet replaced itwith a series of walled realms that don'ttrust each other and are constantly beinginvaded. Each invasion is seen as fatalthanks to the Maginot line mentality.

EV6 rides very well over the V4 network(modulo NATs) though native implemen-tations would be nice. V4 tunneling doeshave its issues. The reason we don'thave V6 is that it is a repeat of my expe-rience with home network. I am notaround to harass the network people tomake sure they get rid of excess baggageand recognize a “just-do-it” mentality as

a priority. This includes being willing todo whatever it takes even if that meansusing TCP tunnels through recalcitrantNATs until they can be brought into line.Efficiency without connectivity is a formof death. It's also vital to address other is-sues like working with the existing V4DNS and extended it with dynamicallyfinding local end points.

And, of course, the edge users must haveencryption because of all those bell headslurking around trying to impose theirterms of service and other forms ofsmart-assed meddling.

Converting the Internet to V6 is a badidea. It must be adopted at the edges tomeet needs and can coexist just fine withthe old Internet. New apps can thrive. Itprobably makes sense to have someV4<=>V6 http but for the most part youwant to go to a v6 server you must havea v6 client

In your newsletter it is vital to distinguishbetween EV6 and BV6. Backbone V6 isnot at all interesting to me though someof your readers may care. As long as BV6is an opportunity to bring back deadideas, however, it may actually make thenet worse. Steve Deering agrees that theideas are bad though I don't know if he'llagree they make the net worse. In shortthe lack of EEV6 (End/Encrypted) V6 isa definite liability for those of us whowant an end-to-end edge controlled net-work that I open to innovation.

David P Reed: Bob doesn't need me toagree with him. His points above aredead on, and in many ways put more suc-cinctly than I would be able to.

Needed: an End-to-EndOverlay

My only amplification would be thatthose of us who see no need for BV6 or atwisty convoluted web of walled gardenswith trolls at all the gates may need to re-volt, and do an end-to-end overlay net-work of our own (just as the original In-ternet was an end-to-end overlay net-work). When is a revolution necessary?When the current market leaders keepbuilding instruments of control -- likeNATs, like usage policies that bar certain

The COOK Report on Internet March 2003

13

kinds of uses, and like attempts to chargemerchants a percentage on every transac-tion on the net -- rather than what theusers really want.

COOK Report: How can the users at theedge install their own IPv6? How do theusers revolt? What is the cost of doingso? What can users do on their own?What do they have to get an ISP to do?

Frankston: Let me start with the ideal-ized answer -- the one I believe would bethe case were I still at Microsoft. It's thesame as the one for home networking. Itwould "just work". In fact, there is a V6implementation in XP right now but it ismissing key "just work" elements:

(1) It doesn't try to "just work" and youhave to do all sorts of setting and tweak-ing. The .Net server version doesn't seemmuch better. (2) It requires a cooperatingNAT and ISP for type 41 packets. Itshould support various alternatives evenif they represent large performance hits.(3) It doesn't do encryption well or auto-matically or universally and confuses en-cryption with authentication. (4) It does-n't allow me to simply use my V4 DNS.(5) It's concerned about transition insteadof simply giving me V6 capabilities.

None of this is fundamental.

To oversimplify, V6 is simply a 128 bitaddress. It is typically divided into a pre-fix portion and a local portion. One ofthese is reserved for a particular way ofusing the V4 address as a prefix but it'spossible to define additional approaches.The local portion is arbitrary and can bebased on the MAC address, for example.

There are some difficulties:

The tunneling packets are a different typeso may run into barriers. There is no rea-son for this to be necessary. Since it is adifferent type of packet you can't just tellyour NAT to forward the packets to agiven port. There lots of ways to name amachine and you can have global prefix-es and local only. Making sure you havethe right one for the return path is a chal-lenge. It's not clear why this has to be aproblem but there seems to an effort tohave special addresses. One reason is to

keep the local addresses constant even ifyour ISP connection flaps but you thenhave to make sure you know the purposeof each address in order to expose appro-priate. I don't see that this should be a bigissue but it adds complexity in explainingthings to users and forestalls "just works"unnecessarily. Encryption without pre-arrangement. I presume it is very doablebut haven't drilled down. Local and dy-namic DNS -- not sure if the protocolsare in place

But none of this is a killer. Roll yourown? Sure. The problem is getting some-one with the right time and expertise todo it. In the MS world everyone is wait-ing (and waiting) for Microsoft.Linux/Unix? There’s no telling.

I think the first step is to simply get thebackbone and edge agendas separate andthen we can make some progress. Toofew people understand the Internet andjust treat it as a telephone-like shoppingnetwork. There is no pain because legacyapps work and few people see beyond thecapabilities of that past. The one middleground is to write apps for V6 and thengo over V4 as an accommodation insteadof thinking "V4"

Farooq Hussain: I certainly have noissue at all - rather I'm in agreement -about making a distinction between thevalue of deploying v6 at the Edge (Ev6)and the deploying it in the backbone(Bv6). There are definitely reasonablerationales for placing emphasis on theedge deployment and good reasons tostay away from Bv6. But assuming thatEv6 gathers some impulsion in the yearsahead, Backbones will have to route v6support http v6/v4 and v4 to v6 and viceversa aside from the necessity for dualstack DNS.

The points that I was trying to emphasizewith Gordon were that:

(1) v6 should stop being viewed as a re-placement for v4 particularly on thegrounds that there will be address spaceexhaustion.

(2) Having 2 IPprotocols on top of all therest that happened to backbones is bothpainful and inelegant, but not necessarily

commercially or operationally unwork-able. Just at the moment, its really hardfor me to see what would commerciallymotivate any carrier to deploy v6 otherthan sensible forward positioning at avery low level of commitment.

(3) I certainly agree with the commentsof David and Bob that the Edge has morevalue, rationale, and probable commer-cial viability than Bv6 at this point. Stillwe'll need to understand the kind of com-mitment that Microsoft is willing to putbehind v6 as well as others like Sony forexample. I'm not sure that I grasp the pic-ture of the grass roots 'brush fire" thatmight accelerate the promulgation ofEv6. I fear that Ev6 is caught in the samegrip of conflicting government policy,giant multi-national enterprise, and otherinstitutional interests. If Ev6 can make acompelling case for adoption based on abusiness/commercial rationale, this -based on the points David has made ear-lier - applies to a different segment of themarket than that of the backbone carriersthough they will be impacted by it.

Frankston: Just to avoid ambiguity --the E in EV6 means EdgeV6 or Encrypt-ed Edge V6. I presume the governmentpolicy question is about Encryption and,at this point, I would make a strong casethat an enlightened government wouldpush strongly in favor encryption -- werewe to have an enlightened government.Encryption is merely like expecting peo-ple to take responsibility and lock theirown doors rather than saying only policeare allowed to have keys.

Role of Microsoft

The reality is that much of this dependson an enlightened Microsoft and, to bevery very specific, Christian Huitema.He's doing everything "just right" but Ihaven't seen a sign of his going beyondthat and separating edge from backbonerequirements and emphasizing encryp-tion and edge deployment.

When I worked there well before Christ-ian joined the company there was noawareness about making Internet connec-tivity simple and I see that as my mainaccomplishment. Nevertheless, it's still avery difficult idea to remain faithful to.

COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA

14

Consumers never asked for home net-working and they are not asking for V6.They are asking for higher and higherfirewalls and, as you can see from theMSN ads, they want only bits with goodintentions to be allowed through. Con-sumers are still asking for more and morebroadband for browsing and not for shar-ing.

As to http v6/v4, I would argue that too isentirely edge. Those who care to listen onboth Internets would have dual portedservers. The big public servers like CNNwill do that, especially if they are strong-ly encouraged and the Microsoft IIS andApaches do it automatically. The old V4servers will still be accessible to all sys-tems since I don't expect the V6 systemsto drop V4 client support even if theydon't have a public appearance. I don'tmind the NATs as a legacy accommoda-tion -- the problem is there is no alterna-tive.

Bv6 is then freed to give more routingflexibility. In fact one can run V4 over aV6 backbone or visa versa. It's just an in-ternal design issue. But why bother withBv6? To the extent that Bv6 is a betterroute description than using a native pre-fix, then you might as well use it. But ifthe routing portion is really only used toget you to the a Ev6 connection point,then it's not as vital. We would greatly re-duce the pressure on the address space.

There is one big problem in my nicestory. I don't like the notion of a singlelevel of V4<=>V6 switching. There is apoint to which the V4 prefix takes youfrom where you switch to your local v6router. One can nest additional routinginternally so you can cascade but the bitlayout story isn't as clean. Still, even withthat caveat, we must get started and giveevery system a public presence and, withencryption, bring the meddler-free Inter-net.

Next we can start teaching people that anInternet Access Provider and InternetService Provider are unrelated functionsand just because ATTBI changes its nameto Comcast is no reason to change theiremail address. But that's the dotDNSagenda and more.

COOK Report: On Friday January 16 Ihad Comcast cable modem service in-stalled. Howver my mail comes fromfast.net. Fast Net also does my DNSservice for cookreport and hosts my website. Therefore Fast Net is my (InternetService Provider (isp) and comcast myInternet Access Provider (iap.)

Frankston: For now, let's get EEv6 ship-ping. And Bv6, doesn't matter to me,that's between Farooq and his managerand shouldn't interest us users thoughyour readers might care, maybe toomuch.

COOK Report: Just to be 100% clear.You meant Edge V6 or Encrypted EdgeV6. Right?

The Necessary Tools

Frankston: When Farooq wrote Ev6, Iwanted to make sure we're explicit. Iwould like it to mean Encrypted andEdge but one can talk about them sepa-rately. Encryption of the pipes can bedone independently of the applications.The purpose is to bring back the naivesimplicity that allowed us to assume ourconversations were not being overheard.App-to-app encryption is more appropri-ate when we want strong security be-tween apps that have their own securitymodel. The pipe security is a compro-mise as is TCP(which gives the apps cir-cuits at the price of potentially very longdelays vs. UDP). In this context, Encryp-tion is then just an edge V6 issue.

Dual port boxes? You get them now.Every XP with V6 is dual port. It's justthat the current V6 implementations are apain to configure and use and get past theNATs and you don't have easily usableencryption. But I should've emphasizedthat we do have dual V6/V4 in XP. Maycomplaint is that they have 90% of themechanism there but haven't put the ef-fort in to the take it all the way to "justwork". There is a very strong case forsaying V6 already is deployed but justnot tweaked. And not encouraged. IIS --Microsoft's Web server and Apache is theprimary one on Linux. Get those twoworking smoothly and you have dualsupport everywhere (at least potentially).

As to complexity. The V4 prefix ap-proach means the net works as-is. NativeV6 routing means that the Ev6 machinesuse Bv6 addresses rather than those withv4 prefixes but otherwise the edge appswork the same. The Bv6 infrastructure isthen able to use native V6 routing rules.Farooq would know more about that thanI do. I presume they have more explicitstructure that allows the backend routersto do something better. Although there isthe risk that better might not really bebetter as things like MPLS and QoS areunproven conjectures and, I would argue,bad ideas. But I presume that those canbe shed and V6 will put more knowledgeabout the net into the bits. But, again,that's a mixed bag.

Hussain: I agree that the MPLS is prob-ably a bad idea in this mix - but it mayprove to be a fact of life in many back-bones. My feeling is (also based on Bob'sobservations above) that Ev6 *may*drive Bv6 but doesn't have to. Left ontheir own backbones don't have sufficientrationale or commercial incentive to goto B6.

Francois Menard: Bob's dead on.

I'm trying very hard to get [Canadian]municipalities to implement IPv6 openaccess across municipal FTTH networksso that MPLS doesn't squeeze-in andend-users become required to run PE`s.I'm seeing ISP's provide value addedservices by offering commercial accessto tunneling servers on their premiseswhich bridge to the good old legacy In-ternet. For as long as two serviceproviders across two different municipalFTTH system would want to intercon-nect with IPv6, there would then be aparallel Internet.

This is in my view a (the) killer app forIPv6 ... I'm not sure why it seems so dif-ficult for people to believe in this ... Andif its true in North America, it'll be evenmore so once open access is properlyprovided in Japan.

I guess that, rather than debating thisphilosophically once again, I'd ratherthrow everything I have into regulatoryinterventions, like the one which is goingout tomorrow in the context of a PartVII

The COOK Report on Internet March 2003

15

of CAIP to the CRTC. This is going toresult in IPv6 being deployed. I'm total-ly convinced that it’s going to end thetight control that incumbents have overlast mile DSLand cable modem in Cana-da and which intentionally destroys in-novation through mandating IPv4 anddenying IPv6.

So Where’s MarketPull?

COOK Report: What is to create a mar-ket pull? How do you package and ex-plain and then ignite that?

David Reed: Products. Too many prod-ucts are limited by NATs.

If Ev6 had the property that it automati-cally ran over NATs, which would not behard, it would be adopted by lots of newproducts. Multiplayer video games, CEdevices like Tivo and Replay, home se-curity cameras, VoIP phones using SIP...all of which are fundamentally "edge-to-edge" devices (not edge-to-server) andthe need to traverse NATs is causinghuge customer support and marketingproblems.

Yeah, I know about STUN and MID-COM and UPnP for IGD's, but when allis said and done, Ev6 overlay network-ing would do the job better, and be morestandard, especially since each of thosesolutions are limited in scope (STUN forhome networks, but not corporate ones,MIDCOM for corporate if only the cor-porate firewall people didn't believe thatnew applications are evil, and UPnPIGD's are essentially restricted to Wintelclients with a fig leaf of a Linux imple-mentation).

But I'm not holding my breath for Ev6from anyone focused on communica-tions (operators, vendors to operators,IETF). The communications industryseems not to care about enabling newproducts at the edge. They seem to wantto control and tax any new innovation,strangling it in its crib. None of this "ris-ing tide lifts all" nonsense for them. :-)

I have some hope for the ConsumerElectronics industry and the computer

apps (hardware and software) industryjust doing something like Ev6 for itsown needs, with or without the IETF.Anyone want to form an IEEE commit-tee for edge-to-edge scalable overlaynetwork standards? We could call it"VigorNet" because it would regain thevigor of the original Internet.

Bob Frankston Jan 22: I want to clari-fy the "Xbox" argument.

People claim we don't need EV6 becausewe can also work around each problemon a case- by-case bases. But that case-by-case basis has brought us a mess.Look at how many of the resulting prod-ucts require setting proxy informationand firewall settings and all sorts ofother stuff. Still other applications re-quire external servers to act as relays.

This problem is not only just creepingossification but also the inability to donew things without a lot of arcaneknowledge that locks one into the acci-dental properties of each of these work-arounds. And as the work-arounds festerthe resulting scab is confused as a some-how necessary part of the environment.It is reminiscent of the Heidelberg scarsthat showed that the student was a greatswordsman and, by extension a scholar.Firewalls have become the condoms ofcomputing and NAT's inability to passinteresting traffic makes them into fire-walls. Finally we have the "marketplace"assuming that all bits have intrinsicmeaning, and asking for these filters tobecome omniscient.

All of this works very very well. At leastby comparison with the ancient world ofscribes and quill pens and 1990. We canbrowse and we can download (a terriblyasymmetric word) and we can use thoseold telephones without having t crankthe magneto. How could anything bebetter? After all, doesn't all this changingstuff threaten all that we have? I wouldargue that the answer is no.

Edge V6 as a SubRoutine Library

One way to think about EV6 is as a com-mon subroutine library just like TCP.

TCP gives us those despised circuits butat the application level where they pro-vide some convenience but are still notintrinsic. The advantage of Ev6 is that itleverages the intellectual energy that hasgone into the Internet protocols andgives us a minimal commonality thathappens to parallel the minimalness ofthe basic IPv4 Internet.

I compare Edge v6 with MIME whichbecame the common way to extendemail rather than having to choose be-tween a lot of different ways to transportbinary and multipart messages. We did-n't transition email to MIME, we justmade it available first to those who un-derstood the need and later to those whojust like pretty stuff. The mistake is to tryto transition the existing Internet to V6.The real need is for enabling the applica-tions that don't work well though the ex-isting protocols. Depending on howdeep their V4 assumptions are, we willfind that existing applications can be re-implemented atop V6 with modest ef-fort.

Without V6 we have no synergy for eachnew application we have to make newarrangements to work around each of themyriad problems. However, with en-cryption and the assumption that new ap-plications aren't hopelessly naïve, wecan reposition the firewall as a tempo-rary scab rather than as protection

In building our edge architecture, weshould then go on to complete the pic-ture with "dotDNS" so we can avoidmaking ICANN the ultimate authorityon meaning. Building an Edge IPv6 ar-chitetcure would also subsume much ofP2P. The P2P effort is about two things:(1) -applications and (2) work-arounds.Each P2P effort has its own novel solu-tion to tunneling through the barriers andits own unique way of generating per-sistent handles (names). Mostly these arejust idiosyncratic and poor reinventionsof the common mechanisms and divertefforts from actually doing anything in-teresting.

While I'm a great believer in market-places, I find that they don't automatical-ly give the optimal path between two

COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA

16

points. If we just look at existing appli-cations, then, by definition, they do not"need" v6. Just like we didn't need the In-ternet for incrementally better TV or forbetter faxing.

It's only by accident that we got to expe-rience the web because the Internet hadlowered the barrier to creativity enoughfor Tim Berners-Lee to hack it togetherin his basement office.

IPV6 is similar. Those of us who haveworked with trying to connect things rec-ognize the importance of a consensusthat would allow us to have devices that“just connect” instead of constantly hav-ing to work around impediments. Themarketplace does have this just connectneed but some many IPv4 accretions arestanding in the way that it can't articulateit.

The simplest solution would be a for ahigh profile end point player like Mi-crosoft (more than just like) do to do theright thing. Given that the “IP Stack”business is problematic in the presence ofplayers who bundle theirs in ways thatmake unbundling extremely difficult, aninterim alternative would best be done aspublic spirited project, perhaps by stu-dents or others with a need. It can be im-plemented as an application level librarywritten atop UDP – it doesn’t have to bedeep in the system. The applicationwould then listen on an EPV6 port forTCP and UDP connections that serve asthe IPv6 tunnel from another system.

This outcome would serve as a V6 shimat the application layer. It can let us as-sume V6 while waiting for “official” im-plementations.

What is Holding BackUse and Deployment

The major impediment is a lack of un-derstanding that the Internet is reallyabout simple end-to-end connectivityand the rest is but a detail. But instead ofviewing the net as a future opportunity,we find that there is the normal tendencyto confuse it accidental properties withwhat they could be and because thoseproperties work then we assume that ob-

viously we don’t need anything else.There should be a clamor to bring backsimple connectivity. Yet everyone seemsto be in love with firewalls, NATs andgargoyles of all sorts. All this is com-bined with a menagerie of hobgoblinssuch as QoS, MPLS and cleverness atworking around problems instead ofsolving them. The Edge V6 need is there.But in order to understand the need, peo-ple need to understand the Internet first.

There are existing implementations ofIPV6 on XP but they don’t “just work”.Perhaps writing an application that sim-ply does the configuration would goalong way towards usability. Havingdone that we would still have to addressthe lack of encryption, the inability to getpast recalcitrant NATs, and the inabilityto leverage the IPv4 DNS entries to nameinterior systems as the norm. Unfortu-nately encryption has gotten entangledwith authentication. I'm not a crypto ex-pert but we should be able to have a mod-est level of crypto between two systemsthat don't know each other.

A secondary problem is that the applica-tion support is uneven but I'm not wor-ried about that since it can come later. Iwant to be able to do simple things likehave a VoIP application that just streamsbetween two end points and doesn't usecomplicated protocols. The fact that suchapplications are not ready attests to theimportance of early Edge v6 availabilityso we can work out such problems.These are the technical issues. They mayhave changed greatly in the last yearsince I looked at the stuff!

Deployment

We should build EV6 on the specifica-tions for BV6 to the extent we can. Doingso will give us an extended address struc-ture. In deployment the most importantstep will be to use the IPV4 address as arouting prefix. We may need an addition-al option for a form of routing that is ableto get past older NATs.

This is entirely separate from the ques-tion of an IPV6 backbone – that is only aperformance issue. It would be nice,however, if the NAT boxes could be re-

purposed as V6 router at the edges of thelocal network. That would give us thebiggest performance improvements.

Taking Advantage of V6

The major value of V6 is in allowingusers to connect devices and not just bigiron and web sites. In doing this, the roleof the DNS in providing a stable handlebecomes very important. Not only do wenot need the .com semantics, in this en-deavor we must be assured that thenames are unique and valid basically for-ever. This is all the more reason for cre-ating a TLD (I call it .DNS) that simplyprovides unique identifiers and NSrecords – the pointers to the actual DNSrecords which would be maintained bythe owner of the identifier.

Note that mobile IP seeks to provide astable relationships but it does it at theplumbing layer. Such application levelstability belongs at the application leveland not the network level. Mobile is an-other example of an experiment maskingas a basic protocol and it has contributedto the confusion over IPV6.

COOK Report: Above the Fold for Janu-ary 29, commented - “Convinced thatlarge "enterprise" networks of the futurewill be shaped by the Internet, by ever-increasing needs for security and mobili-ty, and by the convergence of voice anddata, Hewlett-Packard's new networkstrategy is to move more intelligence andcontrol from the core of a network to itsedges, using cheap switches populatingthose edges.” We sent the url to BobFrankston and asked for his evaluation.

Frankston: It is an example of screwingthings up. Notice level four prioritization-- that's a synonym for breaking the end-to-end connectivity for users by secondguessing the applications. And not aword about V6 or extending the address-ing model. But lots about security in thenetwork which means more and morespeed bumps, twisting passages andmeddling police biddies.

Note their press release “Its ProCurve5300 series switches delivered last sum-mer, for example, which cost about $65

The COOK Report on Internet March 2003

17

per 10/100 port, implement a broadrange of security features as well asLayer 3 and Layer 4 traffic prioritizationfeatures. "They let the customer deploynext-generation intelligence at the net-work's edge at a commodity price point,"

said Clark.

This sure looks like another one of thembellheads run amuck. It would be won-derful if they provide more informationabout the network but adding intelli-

gence and discrimination into the heartof the network (the edge is in my PC, notin the IT switch) is just more of the oldtelco control paradigm.

By Bob Frankston and from http://www.Frankston.com/public/ESSAYS/EncryptedIPV6.asp

We can loosely separate two agendas:

The backbone agenda is about improving the efficiency of the Internet infrastructure. For the sake of this essay I will only notethese issues to the extent they seem to interfere with the edge agenda.

The edge agenda is about making more addresses available so each host can have a public presence. It is about making moreaddresses available as well as improved protocols for automatically assigning addresses. For simplicity I'm focusing on the in-crease in the number of addresses. IPV6 can be deployed at the edges of the network using the existing IPV4 network as atransport.

The two agendas are intertwined to the extent that there must be an agreement on the format of an IPV6 packet and the lay-out of the IP address. But now that there is agreement on the packet format, we can and must deploy IPV6 from the edges.

With and Without

To understand the importance of IPV6 we can compare two scenarios.

Without: If we continue business we will simply accept that the Internet used to be exciting but we have to get back to busi-ness as usual. Experiments at public access will have mixed results and all-to-often will fail. Hotels will provide some accessbut it will be limited and expensive. We will find the Internet is increasingly like television with the transport providers care-fully selecting which services will work and how well they will work. To most people this won't seem to be a problem and theeconomic doldrums will seem to be a higher priority. After all, this is the post Internet era and we should reduce our expecta-tions.

With: I'll have to tone this down to be taken seriously. But think about being able to take your computer anywhere and it wouldjust be connected. But why not? Especially if I could just drop an access point anywhere and connect simply and securely.What might not be obvious is that the kind of "Moore's Law" price/performance improvements that have made email free(once one has paid for a pipe to the rest of the Internet) would operate to make these access points act as part of a commongood in the same way that we generally allow others to benefit from porch light or a restaurant doesn't charge for tap water.These aren't free either but it would seem counter-productive to try to charge a passerby who uses that light to read a map. Thekey to driving this cycle is simplicity. This is not the post-Internet era. We haven't even started to explore the possibilities.

One lesson I've learned with VisiCalc is that seemingly minor decisions can make a big difference. In making home network-ing a normal retail product I took a step towards demystifying the Internet and making connectivity just another commodity.But I was only able to take the first step. I accepted the evil of NATs (Network Address Translation) as I awaited the deploy-ment of encrypted IPV6.

We have waited too long and there is no reason to wait any more since IPv6 can be deployed from the edges without waitingfor any changes to the Internet itself!

Two Internet Futures - With EdgeIPv6 and Without Edge IPv6

48

Executive Summary :

The Action IS TakingPlace at the Edge, pp.1-4 Full Article

At the same time that AOL and thephone companies are trying to staunchthe flow of blood (cash) from the center,innovation is taking place at the edges.The US has built a bankrupt nationalfiber system. Under Michael Powell, theFCC zigs and zags faster than a speedingbullet between innovative spectrum pol-icy and a retrograde insistence that, ifjust allowed, the walking dead of lastcenturies telecom, the LECs and CableCos will invest in building meaningfulinfrastructure.

Powell, it seems, isn't much interested ingetting the details correct. Rather thantaking the trouble to understand the dy-namics of the technology in the marketplace as demonstrated by Clay Shirkeyin his ZapMail essaythat is republishedin this issue, Powell goes on to insist thatit was unfair regulation imposed by hisdemocratic predecessors that has bank-rupted the industry.

According to an article in the Februarysecond New York Times http://www.ny-t i m e s . c o m / 2 0 0 3 / 0 2 / 0 2 / b u s i n e s s / y o u r-money/02FCCC.html?ex=1045173242&ei=1&en=c807a35b91f72fd1, Pow-ell has asserted that deregulation "shouldnot be like a dessert that you serve afterpeople have fed on their vegetables andis a reward for the creation of competi-tion." Rather, he said, deregulation is "acritical ingredient to facilitating compe-tition." Powell is talking the same naivefaith in industry self-regulation that putthe ICANN fox in charge of the DNShen house. Powell's statement ignoresthe central issue that the phone compa-nies would be acting against the interestsof their stockholders, when, if given achance, they did not charge the most ex -tortionate rent for the use of their mo-nopoly possible.

The Times writes: "Mr. Powell and his

supporters say a change in the rules willstimulate the economy by encouragingthe largest phone companies and their ri-vals to build more networks and spendmore at equipment makers like Lucent,Corning, Cisco and Intel." If this is whatPowell truly believes, he is living in adream world and ought to be removedfrom his position by the Congress for in-competence. The fact is the phone com-panies cannot under any circumstances,except those of government enforcedmonopoly high prices, use the networksthey already have. Given their debt theyhave no money to buy new equipmentfor new networks. Let's look at theequipment makers that Powell rattles offas companies that would allegedly bene-fit. Cisco yes, Intel perhaps. But Lucenthas gone from 10 billion a year in rev-enue to two billion because it doesn'tmake equipment that sane managementwould buy were it too invest in a newnetwork. With all the fiber in the groundthe only new market in fiber for Corningis fiber to the home and if Powell getshis way and gives the telcos a monopolyon that, no sane homeowner would wantit.

But Powell, it seems, is interested muchmore in ideology than in accurately fig-uring out where the technology is going.Dave Hughes caught Powell giving Sen-ator Brownback of Kansas in correct in-formation about Wi-Fi in his testimonyon January 20 as Powell stated that theradios used would transmit at best 300feet on an 802.11b network.." Hughesskewered Powell in public as well heshould have. Within 24 hours Hughesheard back from a Powell assistant."Thank you for your comments. In thepassage you reference below the Chair-man simply made a mistake." One won-ders how many "mistakes" MichaelPowell is making these days?

The Canadians are not making mistakes.They are building a working nationalfiber system. They are investing 200 mil-lion dollars in linking all public schoolsthroughout Quebec with fiber and are

doing it such away that all municipalgovernments with be on the same fiber.This includes northern Quebec whereonly the most remote villages will relyinstead on broadband radio.

Robert Proulx President of XIT telecomin Quebec told us in a February thirdconversation that his small company hasall the business that it can handle includ-ing major fiber community networkbuilds in Hungary and in Jordan.

The Canadian CRTC is taking a very dif-ferent tack from the American FCC.Telecom in Canada is understood as amajor national infrastructure resource inthe same way the US understood the in-terstate highway system 50 years ago.Now our dominant ideology permits in-vestment only in private corporate re-sources. Already 17th among the globalusers of telecom services according to arecent OECD study, the United Stateseconomy will suffer in coming years be-cause of our current ideological short-sightedness.

The future is in asset-based and cus-tomer-owned networks. We have in-stalled cable modem service in order tomove our long distance calls to Vonage.Suddenly unlimited long distance in theUS is flat rate. All of Canada is 5 centsa minute and most of Europe and muchof Asia is not much more.Innovation atthe edge is possible and as prices contin-ue to fall the huge companies that Pow-ell want to serve will become more andmore unwieldy. Fiber to the home isworth having only if the homeowner cancontrol it.

Meanwhile the edges continue to canni-balize the center - like a million termiteschewing on the soggy log of the PSTN.BellSouth was the first ILEC to ac-knowledge the inevitable and at the endof January announced that it wouldbegin to resell Vonage to its DSL cus-tomers. See http://news.com.com/2100-1033-982606.html And from a trustedsource we are told that in Japan NTT has

The COOK Report on Internet March 2003

49

effectively ceased development in its cir-cuited switched landline network.

Backbone v6 going nowhere pp. 5 - 11Highlights, Full Article

Farooq Hussain explains how reasonableuses for IPv6 in Internet backbones haveevaporated. DHCP and Nats acting asfirewalls have gotten the need for v6 as asources of extra address space well undercontrol. The idea of universal address-ability across the internet for all deviceshas receded in importance.

He cites very interestingly that in theUnited States almost all the support forv6 comes from the defense department.In Europe it comes from the EuropeanCommission and in Japan from a man-date by the Japanese government.

Mandate or not there no longer seems tobe any market pull. End to end applica-tions like voice over IP that once werethought to be dependent on v6 are beingre-engineered to work with NATs in thev4 world.

3GPP, which is the third generation mo-bile project, adopted IPv6 as their proto-col of choice in 1999. In doing so it gavev6 the strongest endorsement it has everhad. Yet because at the height of thebubble the 3GPP people decided to buildtheir own internet parallel to the globalv4 internet, their plans now seem rathersilly. With the slowdown in wirelessgrowth has come a slowdown in wirelessdemand for IP numbers. If we assumethat with the arrival of software definedradios over the next few years radios willcommunicate with each other on thebasis of IP rather than geography this islikely to delay indefinitely the need forv6 in wireless devices.

SONY has announced that all of its de-vices will speak v6. However in the ab-sence of widespread v6 deploymentSONYs products will also have to com-municate in a v4 world, The problem isthat if the communicate well in v4 thereis likely to be no use for their v6 capabil-ities. It seems that even Jun Murai is nolonger promoting v6 wholeheartedly in

Japan. Farooq concludes that we willlikely have the worst of all worlds withmost of the internet running v4 and fewisolated instances of v6.

When we asked Farooq about the Janu-ary 22 announcement by Telehouse of anIPv6 peering exchange in New York, heresponded PAIX has had the ability tosupport v6 for at least two years, Themore interesting announcement was theone for the exclusively v6 exchange setup in France as part of the EC initiative.Also has hardly any takers except thosewho are compelled by politics to go thereby virtue of being participants in the ECinitiative. It's fine for Telehouse to makethis kind of announcement but the capa-bility is of little commercial interest ei-ther to enterprise or service provider net-works. There's simply not enough trafficvolume with v6 and there are so few na-tive v6 networks that none of them needto go to Telehouse or any other IX to ex-change traffic. The IX's used for v6 havebeen established primarily to foster R&Eprojects and are sustained primarily onnon-commercial rationales.

Edge based v6, pp. 11 -17 Highlights, Full Article

v6 in the backbone. Farooq agrees. Itseems that v6 at the edge can be used byend users to establish their own applica-tions and perhaps even routing by usingv4 addresses in the v6 packet headers.

Frankston has written: IPv4 (or just "IP")represented the birth of the Internet byshifting the power to define the networkto the users at the edges.

The Internet has thrived because supplyis driven by demand. New applicationservices are supported by simply provid-ing more transport (or IP) capacity.Rather than wait for new capabilities tobe defined, users will create their ownsolutions. (When I say "users" I don'tmean all users create applications. It onlytakes one motivated, creative individualwith some time on their hands to createan application that will be adopted bymillions of others. We just don't knowwhich user that will be.)

Francois Menard commented: I'm tryingvery hard to get [Canadian] municipali-ties to implement IPv6 open accessacross municipal FTTH networks so thatMPLS doesn't squeeze-in and end-usersbecome required to run PE`s. I'm seeingISP's provide value added services by of-fering commercial access to tunnelingservers on their premises which bridge tothe good old legacy Internet. For as longas two service providers across two dif-ferent municipal FTTH system wouldwant to interconnect with IPv6, therewould then be a parallel Internet.

A current problem is the absence of agood v6 tool set for end users. Right nowit is not clear where one will come from.Standards development would proveuseful. But by whom? The IETF is veryunlikely. The IEEE perhaps. The Con-sumer Electronics Association claims tobe doing work in the area. Unfortunately,we have not had enough contact to eval-uate them.

ZapMail pp.18 -20Highlights, Full Article

When does a service become just anoth-er product that the phone company’s cus-tomers can deliver best and at lowest costfor themselves?

Clay Shirkey has written a powerfulessay that likens the Local ExchangeCarriers’ world view to that of Fed-Exwhen it though it needed to build a faxnetwork to gain an advantage that otherovernight carriers didn't have to offertheir customers only to find that the cus-tomer could deliver information by faxmuch more cost effectively themselves.

The business Fred Smith imagined beingin -- build a network that's cheap to runbut charge customers as it if were expen-sive -- is the business the telephone com-panies are in today. They are selling us akind of ZapPhone service, where they'vedigitized their entire network up to thelast mile, but are still charging the highand confusing rates established when thenetwork was analog.

COOK Network Consultants, 431 Greenway Ave. Ewing, NJ 08618 USA

50

Discussion of ZapMail,pp. 21-24 Highlights, Full A r t i c l e

Adrew Odlyzko agrees that VoIP willlead to the flat rate commoditization oflong distance phone service. but hethem wonders about wireless being ableto create a substantial enough infrastruc-ture for voice communication.

A product known as Locustworld mayhave the answer. "A UK company hasproduced Mesh wireless technologywhich you can buy and install, today, forunder £300. Fancy setting up as a rival toBT Openworld? Even in a remote vil-lage? Easy: buy a Locustworld Mesh-Box; half the price of a home PC. You'rein business."

"The software is the key to Locustworld.Written by text-message pioneer Jon An-derson, it configures a group of wirelessaccess points into a coherent "mesh" andconnects them to any broadband Internetnode available."

"Most experts regard the mesh approachas hugely complex, because of the effortneeded to set up the mesh. The systemused to be known as a "parasitic net-work" - although the fashionable termthese days is "symbiotic" - the idea isthat you turn a group of wireless nodesloose, and tell them to introduce them-selves to each other. Then you set uproutes through the mesh. It can befiendishly complex, but Locustworld'smesh does this for you. You just buy thenode from them: the current model is£250 plus VAT."

"The last legal obstacle, according tofounder Richard Lander, was the deci-sion by Oftel, allowing people to sharetheir broadband with up to 20 others.

Farber Faulhaber versus Open Spectrumpp. 25 - 30 Highlights, Full

Article

A discussion of the problems created bythe presentation of a paper that looks tothe past encourages spectrum auctions

and their maintenance of property rightsin spectrum while saying oh by the wayyou open spectrum folk may be given socalled "easements" since your magicalradio technologies will not get in theway of our much more rational corporateapproach.

How NSF WasPrevented fromRemoving the govern-ment from domainnames? pp. 31 -34Highlights, Full Article

Don Mitchell explains that a major poli-cy change by the National Science foun-dation was aborted. The change wouldhave ended government involvement inthe DNS. It could have nipped ICANNin the bud. But this was not to be..

Had the cooperative agreement conclud-ed in spring of 1997, as the NSF intend-ed, the problem of institutionalizing theIANA function would have been forcedout on an open table (or, possibly mademoot) by the demand for (and creationof) additional TLDs. It might also havebeen forced into the courts. It certainlywould have become more clear to manymore people that one of the most criticalunderpinnings of the Internet, the IANAfunction, had no basis in law. Neitherdomestic nor international. If the playhad been open, the high stakes maniathat festered into the Internet bubblemight well have not reached such a feverpitch. The industry might not have rid-den so high and fallen so hard.

The over ruling of NSF plans for termi-nation by Burr and her ISOC clique andthe resulting extension of that agreementallowed a small number of high stakesplayers to keep the game closed. Thegame was still closed in June of 1999when in the ICANN board emails wepublished Esther Dyson, IBM, Vint Cerfand Mike Roberts hatched a strategy toget money for ICANN from the venturecapitalists of Sand Hill Road by warningthem that their investment were in dan-ger if ICANN did not succeed and bymeeting with Tom Kalil in the WhiteHouse to seek support. Today the in-

vestments of the Sand Hill VCs havelargely vanished, the IANA function isstill not institutionalized. Indeed todayFebruary 3, 2003 the IANAfunction wasjust handed back to the same closedgroup of high stakes players who professto operate ICANN with openness withauthority. In reality the game is stillclosed.

Lessig on Governancepp. 35-40 Highlights, Full

Article

Larry Lessig in the document that fol-lows gives the best overview that wehave seen of the details under girdingICANN’s construction in the year 1998.In the talk that we republish with his per-mission, he shows how the GIP ISOCClique found in Joe Sims an attorneywho enabled them to take advantage oflibertarian distrust of government to cre-ate an ICANN that they could use fortheir own narrow ends and brought onfour years feuding and distrust. ICANNfrom the very beginning was broken.Such was the distrust of government thatno one would own up to seeing the bro-kenness. Lessig saw it however and hisanalysis of what could be expected fromICANN from the position of hindsightmore than four years later reads likeprophecy.

News Item Dave Hughes to Chairman Powell - Jan 22, 2003

FCC Chairman Powell in testimony to theUS Senate. "That's the way that currenttechnology is configured and deployed.Right now the leading standard of 802.11 ab and g in their very first have a limit intheir range. At best 300 feet on an 802.11bnetwork.. " THAT IS AN ABSOLUTELYFALSE STATEMENT!!!!! 'at best 300 feet'WHY DID HE MAKE IT?

1. There are over 10,000,000 Wi-Fi sys-tems out there. 1.5 million more eachmonth. 2. There are over 2,500 and proba-bly over 4,000 Wireless ISPs using Wi-Firadios doing business across the UnitedStates as I speak. Largely RURAL. I willwager not ONE of them is serving cus-tomers 300 feet or less. Most are from 1

51

The COOK Report on InternetCOOK Network Consultants431 Greenway AveEwing. NJ 08618, USA

mile to 10 MILES using off the shelfequipment certified by the FCC and with-in the power limits - 36dBm EIRP - pre-scribed as the maximum for 802.11b ra-dios. 3. Cisco sells tens of thousands of802.11b 'Aironet' radios which are AD-V E RTISED as reaching 18 miles at11mbps or 25 miles at 2mbps!h t t p : / / c i s c o . c o m / u n i v e r c d / c c / t d / d o c / p c a t / 35 0 w l b r.htm#fea 4. Young Designs Incsells COMPLETE 'Wi-Pop In A Box' sys-tem for communities advertised at 12miles! Standard, certified, systems. 5.Well funded companies are ramping up todeploy Wi-Fi across cities all over the US.One announced today it was targeting80% of the Front Range Colorado popula-tion. And yes, they will backhaul overbroadband wired networks, to answer yourquestion accurately, Senator Brownback.6. I have spent the last 3 YEARS buying,deploying, testing Wi-Fi 2.4ghz as well asother Wi-Fi Bands (915mhz, 5.7Ghz) ra-dios for 4 more years from half a mile to15 and more miles. And I AM a WirelessISP ALL of whose customers are Wi-Fi2.4ghz at ranges from a third of a mile to2 miles!

I KNOW what I am talking about. Whydoesn't the Chairman of the FCC? Or hisStaff, who prepared him for this Hearing?Or is there a hidden agenda there? Thatkind of completely false and misleadingstatement before Congress angers me! Forin effect he was telling Senator Brown-back, whose 'colleagues' and constituentsC O R R E C T LY identify Wi-Fi as ONEtechnology which can bridge the 'lastbroadband mile' until whole new genera-tions of radios are invented, that Wi-Fi isof NO REAL VALUE for Broadband. IKNOW what I am talking about. Whydoesn't the Chairman of the FCC? Or hisStaff, who prepared him for this Hearing?Or is there a hidden agenda there?

That kind of completely false and mis-leading statement before Congress angersme! For in effect he was telling SenatorBrownback, whose 'colleagues' and con-stituents CORRECTLY identify Wi-Fi asONE technology which can bridge the'last broadband mile' until whole new gen-erations of radios are invented, that Wi-Fiis of NO REAL VALUE for Broadband.

The Customer Becomes theNetwork - $395 Now avail-able.http://cookreport.com/assetbased.shtml

Subscription RatesChoice of either ascii or Adobe Acrobat(PDF) format 1. Individual; College orUniversity Department; or Library; or SmallCorporation - $300 2. Corporate - (revenues$10 to 200 million a year) - $395 3. LargeCorporate- Revenues of $200 million to $2billion per year - $495 4. Very LargeCorporate- Revenues of more than $2 billionper year - $595

Site License: The right to distribute asciiand PDF via email to all employees of cor-poration. 5. Small corporate: $495 6.Corporate: $775 7. Large Corporate: $10008. Very Large Corporate: $1250 . SiteLicense Distribution via intranet web site$400 a year additional. See www.cookre-port.com for more detail

Gordon Cook, PresidentCOOK Network Consultants431 Greenway AveEwing, NJ 08618, USATelephone & fax (609) 882-2572 Internet: cook@cookreport.comHow to use interactive features of thisPDF. Click here with hand tool.

The COOK Report on Internet March 2003