Interesting People mailing list archives
WORTH READING Google Plans to Build Ultra High-Speed Broadband Networks
From: David Farber <dave () farber net>
Date: Mon, 15 Feb 2010 09:23:04 -0500
I think this is worth on list djf Begin forwarded message: From: ken <ken () new-isp net> Date: February 15, 2010 7:56:05 AM EST To: Bob Frankston <Bob19-0501 () bobf frankston com>, ip <ip () v2 listbox com>, David Farber <dave () farber net> Subject: Re: FW: [IP] re Google Plans to Build Ultra High-Speed Broadband Networks Bob, I fully expect that Dave will opt to take this off-list as it seems that we may be some of the few who are interested in this discussion, which, to be quite blunt with you, may be indicative of why this problem exist in the country. Speaking directly to your inquiry as to whether I accept that "prices and difficulties are due to policy not technology" I would counter by saying that this is not a binary situation. We do agree that policy is a contributing factor in the cost of these networks and I would even go as far as to suggest that policy could be considered to be one of the most detrimental influences in the ongoing delay the construction of next generation communications infrastructure. Where we seem to have a disconnect is that there is a price, an incredibly steep cost, to building one of these networks. Here's a down and dirty analysis of that cost for your consideration. According to the NationalAtlas.gov, there are four million miles of roads and streets in the United States. http://www.nationalatlas.gov/transportation.html The Fiber-to-the-Home Council claims that "Fiber to the home networks now pass more than 15 percent of the homes on the continent..." If we are willing to accept those numbers we have the following data to work with. 4,000,000 X .85 = 3,400,000 miles of raodway that needs to have fiber laid. There are roughly 110,000,000 residential units in the US (including MTUs) - with no allowance for business structures. 85% of 110,000,000 gives us 93,500,000 homes which need to be connected. Using the $5K per mile number (which is certainly low for locations like NYC) we can put a minimal price on what it would cost to connect the remaining unserved part of the country. 3,400,000 X 5,000 = $17,000,000,000 Now let's assign the $600 number quoted to connect each household that is currently not serviced. 93,500,000 X $600 = $56,100,000,000. All total, we have $73,100,000,000 - which in the greater scheme of things is not a lot of money, as we are both likely to agree. But, remember, there is no allowance for trucks, warehouses, offices, and all of the necessary additional supporting material and resources that a project like this would require. If I had to put a number on that I would suggest that the additional costs would more than quadruple the number provided above. But here's the rub... This year the money handed out for heating energy assistance was cut but roughly 50%. In my area of the country, we have serious problems as senior citizens are turning their thermostats down to dangerous levels. Selling the idea that tax dollars should be used to pay for what many Americans consider to be a private corporation's property is going to be a tough sell. I fully expect that you and I would consider this to be an excellent investment. For my part, there are any number of defense department projects I would happily jettison in exchange for this infrastructure but I doubt we could push this through Congress. One other point that you raised, why Brett holds the views that he does, I would not presume to speak for Brett but as someone that invested a lot of his own money to build a WISP I believe I can speak to one aspect of this point. Exactly how do I get compensated for the work and money I laid out, both from a business standpoint and as a way of investing in my community? If the government comes in and uses my tax dollars to put me out of business, what motivation is there for anyone to make any such investment in the future? With that said, forget me - what about the huge corporations who have invested tens of billions of dollars of their investor's money? How does that get settled? Respectfully, Ken DiPietro Ellerslie MD Bob, if you have some free time I would enjoy taking this discussion to the phone, as email only lends itself to less involved subject matter. On 02/14/2010 11:09 AM, Bob Frankston wrote:
Do you mind if I sent this to the OIA list … I’ve got to admit that I’m frustrated by some of Farber’s forwarding decisions but it’s his list and his choices. I may edit the first paragraph to be less IP-centric. *From:* Bob Frankston [mailto:Bob19-0501 () bobf frankston com] *Sent:* Saturday, February 13, 2010 20:18 *To:* 'ken' *Cc:* 'dave () farber net'; 'ip' *Subject:* RE: [IP] re Google Plans to Build Ultra High-Speed Broadband Networks Dave – Before I spend too much time on this I’m curious whether others on this list accept these limitations and how they react to http://rmf.vc/?n=FIT. We seem to be going in circles. Ken – I’m also interested in your reaction to the talk. If you accept that prices and difficulties are due to policy not technology then why do you cite them as reasons for that things can’t change? Google can change the dynamic they can effect policy and challenge the telecom business model and remove the impediments to hypergrowth – in fact we’re poised for rapid catch ups. Shouldn’t we encourage them? I’m puzzled by the comparison with Brett and Google. Google can make this capacity available to all. If Brett weren’t trying so hard to be a subscale ISP he’d benefit greatly by being one of those people users hire to help them with do networking and use services that taking advantage of the infrastructure. They’d get local support rather than depending on the benevolence of a distant provide whose only interest is maximizing ARPU. -----Original Message----- From: ken [mailto:ken () new-isp net] Sent: Saturday, February 13, 2010 18:05 To: Bob Frankston Cc: dave () farber net; 'ip' Subject: Re: [IP] re Google Plans to Build Ultra High-Speed Broadband Networks Bob, I tell my child that everything is easy once you know how to do it. I explain to my mother (89 years of age) how to add attachments to email. While customer service is one component of the operational cost, the hardware necessary to operate a multi-gigabit network does not come cheap, nor do the people who support it. In addition, there is the network itself,and even at a cost of $1,500 per home connected, that money needs to be recouped. While an argument can be made that each household should own their own connection, that money still needs to come from somewhere and I can assure you that where I live the majority of households could not and would not raise that kind of money. This is why there is an associated cost with each service the customer finds valuable. At the risk of stating the obvious, there is also a cost to connect to the POTS network, to deliver content (for television or VoD) and while Google might enjoy peering, most smaller companies pay extremely dearly for their upstream connectivity as Brett has mentioned. I completely agree with you, the technology isn't the holdback, it is an acceptable business model that needs to be developed, one that will be agreeable to just about everyone. Respectfully, Ken DiPietro Ellerslie MD By the way, I applaud the philosophy you employed at Visicalc. In my opinion it is the only way to go. On 02/13/2010 05:44 PM, Bob Frankston wrote:The question is not whether running a network need be expensive asmuch as running a network is the right framing. This is why framing and design are so important. Whenever something was hard to explain in VisiCalc we fixed the program rather than tried to explain it.-----Original Message-----From: ken [mailto:ken () new-isp net]Sent: Saturday, February 13, 2010 16:10To: Bob FrankstonCc: dave () farber net; 'ip'Subject: Re: [IP] re Google Plans to Build Ultra High-Speed BroadbandNetworksBob, (et al)Answering somewhat briefly and without changing context...I absolutely agree with respect for the need to completely reframe thetelecommunications regulatory system and I have high hopes that ChairmanGenachowski will be able to make those changes. Ironically, thesechanges would be good for the telecommunications industry.I will disagree with you in regard to your assertion that running anetwork is not expensive. Perhaps we have different views of whatexpensive means but I can assure you, if it were cheap, every used carsalesperson in the world would be in the business.As far as support (and apologies, in advance for sounding elitist) weseem to take for grated that many of the people we associate with are asintelligent as they are. As someone who owned and operated an WISP, Ican reliably tell you that support is an ongoing issue - most of whichhas nothing to with the network itself. Hell, I supported PCs for almosttwo decades and if I had a nickel for every time I had to explain what adouble-click or a right-click was I'd be giving away gigabitconnectivity today.In the final analysis, I believe we are mostly in agreement.Respectfully,Ken DiPietroEllerslie MDOn 02/13/2010 03:33 PM, Bob Frankston wrote:I do think we agree on much but I want to focus on the crux of theproblem – the telecom framing. This is why I gave a brief talk<http://rmf.vc/?n=FIT> on reframing the problem.*Do you really mean that running a network is intrinsically expensive?*I don't want you (the editorial you and I) to run a network I just wantyou to maintain the facilities I use to do networking. I don't want topay you for networking for me any more than I'd want to be forced to paysomeone for an arranged marriage. The costs you cite for networking area cost of a bad architecture. They are not a necessary cost that Ishould be forced to pay. This is why we need to understand telecom asframing and not as something fundamental. Remember that the Internetdemonstrates that best efforts <http://rmf.vc/?n=UAC@BestEfforts> worksfar better than network services like circuits.We seem agree that many of the choices are constrained by policydecisions driven by an attempt to keep the concept of "telecom" alive asper the 19^th century design point. The very reason we needed aregulatory system is that it's a dysfunctional construct, even if it wasgenerative in 1860.As you note we can increase the capacity of the fiber with betterelectronics but when we did it we got a collapse in the price for fiberand telecom services. So we stopped the improving the technology andunder-deploy it just to keep the current dysfunctional model alive.*And we accept the fairy tale story of scarcity even though we shouldknow that the Hush-a-phone people are still lying to us! *They keeptelling us why nothing can change because empowered user is the enemy ofARPU. They frame customer support in terms of their broken architectureand business model.It's another story meant to scare children and other naïfs. When you doconsumer software (and hardware) you have to address the entirety of theuser experience rather than relying on customer support or, worse,making support a profit center. With home networks my goal was "justwork". Alas, thanks to the problems with today's Internet protocols itdoesn't work as well as I'd like but that has created an industry ofthird party support like Geek Squad.I have a Nexus phone that has its support limits but supported phonesaren't really well supported either because of the gratuitous complexityof the cellular /business/ model.The key is in having an architecture that reduces the need for supportand you do this by decoupling the elements rather than havingall-or-nothing silos.And you learn. In this case Google seems to be doing a limiteddeployment so that it can learn. I hope others learn from theirexperience so that we won't have to depend on Google for infrastructurebut will instead be able to hire local companies to install and maintainthe physical infrastructure.After all, it is far less expensive that complicated things like sewersand sidewalks.Note: VoIP is not a service of a network provider – it's something userscan do using software and by optionally using third party providers.Note 2: If you take that $5K/mile and break it down per strand it's notmuch. If you view the bundle as a whole and break it down per gigabitit's trivial. If you take advantage of flexibility and share paths youreduce the miles, if …. Well, I can go on but you get the point.*From:* Dave Farber [mailto:dave () farber net]*Sent:* Saturday, February 13, 2010 14:26*To:* ip*Subject:* [IP] re Google Plans to Build Ultra High-Speed BroadbandNetworksBegin forwarded message:*From:* ken <ken () new-isp net <mailto:ken () new-isp net>>*Date:* February 13, 2010 1:51:10 PM EST*To:* Bob Frankston <bob2-39 () bobf frankston com<mailto:bob2-39 () bobf frankston com>>*Cc:* dave () farber net <mailto:dave () farber net>, 'ip'<ip () v2 listbox com <mailto:ip () v2 listbox com>>*Subject:* *Re: Google Plans to Build Ultra High-Speed BroadbandNetworks*Bob, (et al)So many points to address...Google Voice is, in fact, framed by the system it is legallyrequired toconform to. If we wish to see real change, this is where thefight mustbegin.Turning to your comment about charging for VoIP, yes, old school is acorrect label for that concept, as I'm sure we both realize that thecost for VoIP service will reach zero in the near future. Youwill notethat the analysis you are pointing to was written in 2005 andeven backthen you will find in my writings that voice would be provided asa freeservice eventually.Certainly cost reductions can be done but you might want to rememberthat running a network is expensive. As David Mercer has correctlypointed out, the customer service alone is an ongoing expense, eventhough it has been my experience that when a well designed isdeployed,customer service is minimized.The $5K/mile refers to a 144 strand bundle of fiber beingtrenched usinga two man team and a Ditch Witch. I am reliably told that Jaguarexpectstheir team to complete one mile of installed fiber per day. I do notknow whether this service is carrying 100Mbps or one gigabit but thecost for hardware to light fiber at gigabit speeds when compared to100baseT is relatively trivial.Let me also raise one more critical factor in this discussion,the PUC(or Public Service Board) can make or break any such deployment.Thereis also the issue of gaining access to the poles, as with aerialdeployments, as many locations have utility poles which have noadditional capacity for fiber to be hung.Respectfully,Ken DiPietroEllerslie MDOn 02/13/2010 11:25 AM, Bob Frankston wrote:As I wrote in http://rmf.vc/?n=BeyondLimits the type ofhypergrowth weassociate with Moore's Law stems from marketplace in which weare veryaccepting of what works. This means that even backhoes aresubject tomajor price-performance improvements because we find ways to getmorebits through existing paths by using better electronicsrather thanbigger trenches.I want to be careful about being overly optimistic – Google'svoiceservices are still framed in terms of carrier concepts likephone callsand the Android store rebates part of the price to yourcarrier butthose can also be ploys to co-opt the incumbents.It's a matter of incentives – carriers need to limit capacity tomaintain the value of their services and their bits. Googleincentivesare aligned with increase usage and capacity. That changeseverythingand upsets the accounting model we see in analyses such ashttp://nextgencommunications.net/blog/2005/05/the-better-model/whichcomingle technology costs with application costs. Note that thatanalysis assigns a monthly charge to VoIP – very old school.This goesright to the points I made in http://rmf.vc/?n=IPTelecomCosts.If Google isn't constrained by carrier architectures it could beverycreative in finding ways to reduce costs and improveperformance. Forexample it needn't build a home-run distributionarchitecture. I canthink of a number of ways to reduce the costs of providingbit pathsamong homes and to reduce the cost of operating such facilities.If itisn't concerned with counting bits it can create a resilientinfrastructure that isn't path-dependent and which scalesless thelinearly with the deployment.My biggest concern is that this still seems to be a pay tosubscribemodel but as this is not Google's main business I hope theycan pastthat. That would allow even further reduction in costs such asserviceshousing complex as a whole and letting the residents divvy it upamongthemselves.If we're talking $5K/mile for example – is that the cost of 1strand or100 strands or bundles? Is that a gigabit or a terabit or a 100terabits? Is it the cost for the backhoe or the glass?Let's encourage Google to experiment. For Google this is a meansand notan end so they have every incentive to drive the process andgive us thefactor of a million improvements we've seen in our home networks.*From:* Dave Farber [mailto:dave () farber net]*Sent:* Saturday, February 13, 2010 09:57*To:* ip*Subject:* [IP] Re: Google Plans to Build Ultra High-SpeedBroadbandNetworksBegin forwarded message:*From:* ken <ken () new-isp net <mailto:ken () new-isp net><mailto:ken () new-isp net>>*Date:* February 13, 2010 9:52:48 AM EST*To:* dave () farber net <mailto:dave () farber net><mailto:dave () farber net>*Subject:* *Re: [IP] Re: Google Plans to Build UltraHigh-SpeedBroadband Networks*Dave,Perhaps these comments and this link will help to shed somelight onthis discussion.George, (et al)Donny Smith (Jaguar Communications) has the lowest connectedcost of anyFTTH network construction company that I am aware of andJaguar trenchesall of their installations.Here is a link to an analysis of Jaguar's network eventhought I amreliably told that their numbers now are closer to $5K/milefor backhaulplus $600 per home connected.http://nextgencommunications.net/blog/2005/05/the-better-model/Now, with respect to whether Google will cherry picklocations, Ibelieve it is only fair to wait and see before passingjudgment.Respectfully,Ken DiPietroEllersile MDOn 02/12/2010 10:54 AM, Dave Farber wrote:Begin forwarded message:*From:* George Ou <George.Ou () digitalsociety org<mailto:George.Ou () digitalsociety org><mailto:George.Ou () digitalsociety org><mailto:George.Ou () digitalsociety org>>*Date:* February 12, 2010 7:31:46 AM EST*To:* "dave () farber net <mailto:dave () farber net><mailto:dave () farber net><mailto:dave () farber net>" <dave () farber net<mailto:dave () farber net><mailto:dave () farber net><mailto:dave () farber net>>, "ChrisSavage () dwt com<mailto:ChrisSavage () dwt com><mailto:ChrisSavage () dwt com><mailto:ChrisSavage () dwt com>" <ChrisSavage () dwt com<mailto:ChrisSavage () dwt com><mailto:ChrisSavage () dwt com><mailto:ChrisSavage () dwt com>>*Subject:* *RE: [IP] Re: Google Plans to BuildUltraHigh-SpeedBroadband Networks*"For example, I recall that Loma Linda (amunicipal fiberbuild) usedsome UK-based inexpensive fiber-laying system usingmicro-trenches,and fiber in some special sheathing to make itespeciallyeasy torun. A completely new network might be able to usesuch asystem in away that an established network provider might notfindattractive."I guess you think that Verizon must be reallystupid fordoing thingsthe old fashion bell-head way and spending $800per homepassed plusanother $800 to hook up an actual subscriber. Thoseyoungwizards atGoogle will just figure out how to cut costs in half.Now let us return to reality. Verizon's costs are thelowest in theindustry because most of their cabling is aerial.Underground plantcosts about 7 times more money and this is one of themajorreasonsQwest is avoiding fiber because 3 quarters of theirhomes useunderground plant while Verizon has the oppositeratios.Google is going to cherry pick fewer than 1% of homesthatwill be thecheapest/shortest aerial runs and communities withthe leastonerousregulations. Then they're going to turn aroundand claimthat this issomehow relevant to the remaining 99% of the nation.George Ou*From:* David Farber [mailto:dave () farber net]*Sent:* Thursday, February 11, 2010 5:43 AM*To:* ip*Subject:* [IP] Re: Google Plans to Build UltraHigh-SpeedBroadbandNetworksBegin forwarded message:*From: *"Savage, Christopher" <<mailto:ChrisSavage () dwt com>ChrisSavage () dwt com<mailto:ChrisSavage () dwt com>>*Date: *February 11, 2010 8:27:39 AM EST*To: *< <mailto:dave () farber net>dave () farber net<mailto:dave () farber net>>*Subject: RE: [IP] Re: Google Plans to Build UltraHigh-SpeedBroadband Networks*Dave,I'd add the following to Chuck's points. Supposegoes big andends up serving 500,000 people. At 2.5people/household that's200,000 customers. I don't have the numbers handybut Isuspect anetwork with 200,000 customers would put them in thetop 20networkoperators in the country. (Numbers fall off prettysteeplyafterVerizon-AT&T-Qwest-Century-Comcast-TW-Cox-Cablevision-Charter-BrightHouse…)Their experiment may be the worst flop of alltime, butwhatever theylearn will reasonably apply to a very large segmentof thepopulace.For example, I recall that Loma Linda (a municipalfiberbuild) usedsome UK-based inexpensive fiber-laying system usingmicro-trenches,and fiber in some special sheathing to make itespeciallyeasy torun. A completely new network might be able to usesuch asystem in away that an established network provider might notfindattractive. If that process was cheap enough, andscaledwell, thatwould be very interesting information for theindustry as awhole (notto say regulators) to see. (Cf. /The Innovator'sDilemma/).It wouldimply that Chuck's back-of-the-envelope costestimates arehigh by anontrivial factor.As another example, I have heard some strong networkneutralityproponents argue that over a reasonable planningperiod, thecost ofadding bandwidth to deal with the demands of the top5% or1% or 0.1%of users who send/receive massive amounts of data isactually cheaperthan the cost of deploying the systems needed tomonitor, limit,and/or bill for their usage. This has always struckme as aninteresting, if a bit implausible andcounterintuitive,assertion. A1 Gbps network might well provide a test of it. (I amreminded hereof the survival approach of the 17-year cicadas thatwe gethere inthe mid-Atlantic. It's called "predator satiation."Withbillions ofdefenseless cicadas available, predators eat allof themthey want,and then get sick of them and mainly leave them alone– withbillionsstill left. Perhaps with currently available appstherereally is anupper limit to how much bandwidth any one person willuse,i.e., maybeit is possible to simply satiate the bandwidth"hogs".)I have to say – win, lose, or draw, Google's proposalhereis one ofthe most */interesting /*things to happen in thebusinessfor quitesome time…Chris S.------------------------------------------------------------------------*From:* David Farber [mailto:dave () farber net]*Sent:* Thursday, February 11, 2010 7:56 AM*To:* ip*Subject:* [IP] Re: Google Plans to Build UltraHigh-SpeedBroadbandNetworksBegin forwarded message:*From: *"Charles Jackson" <<mailto:clj () jacksons net>clj () jacksons net<mailto:clj () jacksons net>>*Date: *February 10, 2010 9:47:32 PM EST*To: *"'Faulhaber, Gerald'" <<mailto:faulhabe () wharton upenn edu>faulhabe () wharton upenn edu<mailto:faulhabe () wharton upenn edu>>*Cc: *"'David Farber'" <<mailto:dave () farber net>dave () farber net<mailto:dave () farber net>>*Subject: RE: [IP] Google Plans to Build UltraHigh-SpeedBroadbandNetworks*Well, Google never claimed to be providing a realbroadbanddistribution service or significant infrastructure.Rather,theyproposed an experiment. On Google's blog, they claimthatthey arelooking to offer service to between 50,000 and 500,000people. If weassume 2.5 people per household, this works out to 20to 200 Khouseholds. If we assume $1,000/passing, $1,000more peractive drop,and 10% penetration, then passing 50 K people (20Ksubs)would costthem $22 million. I think they can afford that.Heck, they might learn enough about future or emergingconsumer needsthat this experiment will be well worth the money.See<http://googleblog.blogspot.com/>http://googleblog.blogspot.com/and<http://www.google.com/appserve/fiberrfi>http://www.google.com/appserve/fiberrfi.There's also the political side. If they pick amedium sizedcommunity (100 K pop), with above ground utilities(easy tobuild),outside the snow and hurricane zones (nointerruption ofconstruction), and a poor cable TV system and noFIOS, theirservicecould easily come out looking golden.Making the experimental network "open" probably coststhemlittle andgives them another political plus.Chuck*From:* Faulhaber, Gerald[mailto:faulhabe () wharton upenn edu]*Sent:* Wednesday, February 10, 2010 9:15 PM*To:* Charles Jackson*Cc:* 'David Farber'*Subject:* RE: [IP] Google Plans to Build UltraHigh-SpeedBroadbandNetworksWell, if we have a charitable organization willing tolosemoney onbroadband distribution, fantastic! We'll let searchadvertisingsubsidize infrastructure. I'm all for that; butthen Idon't own anyGoogle stock.My point re: CLECs is that a firm actually needs tohave realexperience in local distribution networks to makethem work(i.e., runwith reasonable reliability, not break down due toweatherand pooroutside plant, not have the fiber chewed up besquirrels,not haverepeaters used for target practice, all the boringstuffthat networkguys know and Silicon Valley guys don't), not justmoney tothrow atthe problem. When they've successfully trenched5,000 milesof fiberunder city and suburban streets and it actuallyoperates fora yearwithout failure, then they might have some networkcred.Professor Emeritus Gerald Faulhaber<http://assets.wharton.upenn.edu/~faulhabe>Business and Public Policy Dept.Wharton School, University of PennsylvaniaPhiladelphia, PA 19104/Professor Emeritus of Law//University of Pennsylvania/*From:* Charles Jackson [mailto:clj () jacksons net]*Sent:* Wednesday, February 10, 2010 8:56 PM*To:* Faulhaber, Gerald*Cc:* 'David Farber'*Subject:* RE: [IP] Google Plans to Build UltraHigh-SpeedBroadbandNetworksI'm game to sharing this exchange with the IPaudience.My response to Gerry's response is that the propermodel forGoogle'sexperiment is not a CLEC (funded by investors tryingto gettheirmoney back) but more like "Green Acres" in whichOliver WendellDouglas can get by even if he doesn't make any moneyfarming. IfGoogle is willing to lose a little money (in Googleterms)and puts agood manager on the project, they can providefirst-rateservice. Ifthey choose a market that is currentlyunderserved, theycould end uplooking pretty good.If they offered service to 100,000 people, thatwould beabout 40Khouseholds. If they got 10% penetration, that'sonly 4Kcustomers.It doesn't take a lot of resources to give goodservice to 4Kcustomers—especially if you are willing to lose $2for every$1 billed.Chuck======================Charles L. Jackson301 656 8716 desk phone888 469 0805 fax301 775 1023 mobilePO Box 221Port Tobacco, MD 20677------------------------------------------------------------------------*From:* Faulhaber, Gerald[mailto:faulhabe () wharton upenn edu]*Sent:* Wednesday, February 10, 2010 8:05 PM*To:* Charles Jackson*Cc:* David Farber*Subject:* RE: [IP] Google Plans to Build UltraHigh-SpeedBroadbandNetworksChuck [Dave, I'm happy to have this exchange on IP,if Chuckagrees]--I mis-spoke. I meant experience in running a localdistributionnetwork, and was a bit sloppy in not beingspecific. Butrunning along-haul network is worlds apart from running a localdistributionnetwork. Evidence? Many of the CLECs /circa/ 2000were run byredundant AT&T operations guys, who thought theyunderstoodnetworks.Turns out they were clueless when it came to localdistribution, andmost went belly-up (helped along by recalcitrantILECs ofcourse).But these guys went into a business they didn'tunderstand whilethinking they did understand it. Running Google'sCDN is noexperience for local distribution.Professor Emeritus Gerald Faulhaber<http://assets.wharton.upenn.edu/~faulhabe>Business and Public Policy Dept.Wharton School, University of PennsylvaniaPhiladelphia, PA 19104/Professor Emeritus of Law//University of Pennsylvania/*From:* Charles Jackson [mailto:clj () jacksons net]*Sent:* Wednesday, February 10, 2010 7:47 PM*To:* <mailto:dave () farber net>dave () farber net<mailto:dave () farber net>*Cc:* Faulhaber, Gerald*Subject:* RE: [IP] Google Plans to Build UltraHigh-SpeedBroadbandNetworksGerry wrote:Google has never run a carrier-grade local networkingbusiness(besides its trivial Mountainside, CA WiFi network)and has zeroexperience in networking. Networking is a verydifferentbusinessfrom anything Google has done before . . .Google's Internet backbone appears to be the secondor thirdbiggestbackbone. Google runs an ENORMOUS network. A recentpresentationstated that Google accounts for about 5% of Internettraffic—behindonly Level 3 and Global Crossing.Seehttp://www.nanog.org/meetings/nanog47/presentations/Monday/Labovitz_ObserveReport_N47_Mon.pdf.Google may not be an ILEC or cable company, but theorganization mustpossess a significant (enormous?) amount of networkingknowledge.They haven't been doing access networks—but there areprobably veryfew entities in the world that spend more on routers.If Google chooses to offer service in a communitythat has apoorcable company and no FiOS, they should find it easyto lookgolden(assuming that they are willing to lose a few hundreddollars perhousehold passed.)Chuck(Charles L. Jackson)------------------------------------------------------------------------*From:* Dave Farber [mailto:dave () farber net]*Sent:* Wednesday, February 10, 2010 6:50 PM*To:* ip*Subject:* [IP] Google Plans to Build Ultra High-SpeedBroadband NetworksI agree with Gerry. DjfBegin forwarded message:*From:* Gerry Faulhaber <<mailto:gerry-faulhaber () mchsi com>gerry-faulhaber () mchsi com<mailto:gerry-faulhaber () mchsi com>>*Date:* February 10, 2010 5:59:59 PM EST*To:* <mailto:dave () farber net>dave () farber net<mailto:dave () farber net>*Subject:* *Google Plans to Build Ultra High-SpeedBroadband Networks*Dave [for IP]I am a huge enthusiast for more broadbandcompetition andwelcomeGoogle into the business. I have always wonderedwhy Google(whose market cap = $179B compared to total UScableindustry =$95B) whined incessantly about the domestic BBproviderswhen itcould well have entered the market itself. Itcertainlyhas thefinancial strength to do so, and has for quitesome time.Itsentry (which this announcement perhaps heralds) islongoverdue,in my book. But this is merely a blogannouncement, andtalk ischeap. Let's be cautious about how much we readinto this.But let's be serious; Google has never run acarrier-grade localnetworking business (besides its trivialMountainside, CAWiFinetwork) and has zero experience in networking.Networking is avery different business from anything Googlehas donebefore, andmy guess is that unless they are in for thelong haul,they willget their head handed to them...by customerswho areunwilling totolerate poorly performing networks. They havealso shownthemselves cack-handed at dealing with thepolitics of localdistribution. Remember the Google/Earthlink SanFrancisco FreeWiFi network proposal?Google, I certainly encourage you to get into thisbusiness. Butthis ain't no search engine biz; runningcarrier-gradenetworksfor commercial and residential customers istough anddemandingand presents challenges you have never encounteredbefore. I hopeyou are up to it.Professor Emeritus Gerald FaulhaberWharton School and Law School, University ofPennsylvania----- Original Message -----*From:* Dave Farber <mailto:dave () farber net>*To:* ip <mailto:ip () v2 listbox com>*Sent:* Wednesday, February 10, 2010 1:25 PM*Subject:* [IP] WSJ TECHNOLOGY ALERT: GooglePlans toBuildUltra High-Speed Broadband NetworksBegin forwarded message:*From:* "charles.brownstein" <<mailto:charles.brownstein () verizon net>charles.brownstein () verizon net<mailto:charles.brownstein () verizon net>>*Date:* February 10, 2010 1:18:05 PM EST*To:* David Farber <<mailto:dave () farber net>dave () farber net<mailto:dave () farber net>>*Subject:* *Fwd: WSJ TECHNOLOGY ALERT:Plans toBuild Ultra High-Speed Broadband Networks*__________________________________Technology Alertfrom The Wall Street JournalGoogle plans to build and testbroadbandnetworksthat could deliver speeds morethan 100 timesfaster than what most Americansuse. Theplan,announced on a company blog, couldexpandGoogle'sposition on the Internet byansweringconsumerdemands for ever-fasterconnections.<http://online.wsj.com/?mod=djemalertTECH>http://online.wsj.com/?mod=djemalertTECHArchives<https://www.listbox.com/member/archive/247/=now><https://www.listbox.com/member/archive/rss/247/><http://www.listbox.com/>Archives<https://www.listbox.com/member/archive/247/=now><https://www.listbox.com/member/archive/rss/247/><http://www.listbox.com/>Archives<https://www.listbox.com/member/archive/247/=now><https://www.listbox.com/member/archive/rss/247/><http://www.listbox.com>Archives<https://www.listbox.com/member/archive/247/=now><https://www.listbox.com/member/archive/rss/247/><http://www.listbox.com>Archives <https://www.listbox.com/member/archive/247/=now><https://www.listbox.com/member/archive/rss/247/>[Powered byListbox]<http://www.listbox.com>Archives <https://www.listbox.com/member/archive/247/=now><https://www.listbox.com/member/archive/rss/247/><http://www.listbox.com>Archives <https://www.listbox.com/member/archive/247/=now><https://www.listbox.com/member/archive/rss/247/><http://www.listbox.com>
------------------------------------------- Archives: https://www.listbox.com/member/archive/247/=now RSS Feed: https://www.listbox.com/member/archive/rss/247/ Powered by Listbox: http://www.listbox.com
Current thread:
- WORTH READING Google Plans to Build Ultra High-Speed Broadband Networks David Farber (Feb 15)