Interesting People mailing list archives
Re: There is no business case for broadband Internet [Was re: Interconnect Scaling]
From: David Farber <dave () farber net>
Date: Sun, 1 Apr 2007 09:26:00 -0400
Begin forwarded message: From: Max Tulyev <president () ukraine su> Date: April 1, 2007 10:51:53 AM EDT To: Bob Frankston <Bob2-19-0501 () bobf frankston com>Cc: dave () farber net, ip () v2 listbox com, dewayne () warpspeed com, dpreed () reed com Subject: Re: [IP] Re: There is no business case for broadband Internet [Was re: Interconnect Scaling]
Hi All, At first, really let's see what we are selling as the Broadband Access? Connectivity? I buy a megabit of transit for $50 (just because of there is still a lack of competition, and we are expecting $20 till the end of this year). Local connectivity (peerings, IXes) is ~$100/month per gigabit. So I can do 1:2 or even 1:1 ratio with reliable price. No, we don't sell connectivity. We do different tariffs for users only to enlarge ARPU (reach people pays more), but not for our connectivity engineering. Wire/spectre infrastructure? Yes, we should include $1-$2 in monthly fee or our customers for maintaining our infrastructure, but it is not main costs. SUPPORT, Yes, THAT'S main thing broadband providers sell. This means customers can call to our suport center at 3am and ask a various question from how to set-up a Windows connection till asking to change (or make, if new one) wire from our switch to their home. Yes. We are selling a support and service. Internet, connectivity and so is a really secondary thing ;) Bob Frankston wrote:
/I really should do this as a paper in its own right but this debate provides useful context./Yes – what is this “broadband Internet”? Are we talking about a TV channel?We need to be very careful about our terms and definitions. The questionis what are we selling and can we keep the price above cost. We alsoneed to be careful about accounting choices in assigning costs. I would argue that the business of selling transport as a service doesn’t scale because it makes far more sense to buy facilities and pay for trenchingand maintenance Max is correct in pointing out that the cost of the physical infrastructure is relatively low – especially if we go wireless. The question is what is the product? If we use the example of the network within the house there really isn’t a cost for the bits – once you’vepaid for a router then the bits themselves have no cost. There is a costfor the wires but that’s a fixed cost. If we scale this outside the home then maintenance becomes an explicit cost but the problem comes when we try to associate the costs with “bandwidth” or bits. As we see in the example of the home network the cost is based on the wires and the devices but not with the bits. With fiber the cost is primarily about the gear we use to light the network not the physical transport (one can argue about CAT-6 vs 3 but that’s not significant for this purpose). To test the rationale for bandwidth pricing we should start with the home network and it would be hard to justify charging people more forgigabits vs megabits. The problem becomes obvious as soon as you run video.If we extend this to the neighborhood one can argue that there is scarcecapacity but that’s a social policy not a technical policy. Far better to deploy sufficient capacity so that the normal applications that use only a small percentage of the capacity won’t have a problem and thosewho do use a huge portion may be affected. Again, we can use local videoas a test of the rational for bandwidth or bit pricing. It would swamp other applications and VoD demonstrates that we already have the capacity for video on a common network. People do want share.Note the mention of xDSL. If you are using wires from a third party thenyou may have to put up with the limitations. But the xDSL technologies do not mean that you can’t do better if you owned the wire. I rememberfirst hearing the term xDSL when working with Tut systems for phone wirenetworking. Marty Graham made the point that the limits are not intrinsic, it’s similar to the point David Reed made about wirelesscapacity. I need to be careful – it’s not necessarily that the capacity is infinite but we can do far better if we didn’t presuppose the limitsand focused on opportunity rather than doling out scarcity. So that leaves the cost of spectrum and peering.While I argue that we shouldn’t have the concept of licensed spectrum wecurrently do and it’s a matter of pricing. There should be significantreduction in cost by avoiding using the wires but the policy frustratesgetting the benefit outside of the unlicensed spectrum. Given the ability to get gigabits using wires and megabits using wireless then charging for bandwidth makes sense only if the cost of investing in wires doesn’t give a huge advantage in pricing and capacity.Both spectrum and xDSL represent arbitrary limits on capacity. There is no real cost for the bits but we are trying to dole out scarcity against a presumed limit. These limits are essential for maintaining prices. Ifwe don't have such limits and there is abundant capacity then the bandwidth pricing model collapses.There is still a business model for deploying and maintaining the fixedassets. If we have the model that this is a community then the company that does bandwidth pricing can also simply deploy as much physical capacity as the community wants to pay for. My argument is that the community’s buying power would reduce the cost to far below what theywould pay with bandwidth pricing. The problem today is that they are notnormally given the option.More to the point the tendency to view the network in terms of services as we do with broadband frustrates rational debate and an understandingof the true costs and opportunities.I do appreciate that Max is offering pure bit transport. Ideally he’s inthe position to switch business model to facilities management. This would be far more difficult if he were in the triple play business and worried about being bypassed. The facilities management business is a very price sensitive business because it is too easy to compare prices when one is not locked into silos but it’s no different than garbage collection or other such businesses. The important point is that this is not really broadband as we know it today when I have to use a FiOSTV router to get their VoD but doing so breaks the home network due to implementation issues of their NAT and other design choices. The remaining problem is the cost of peering. It would be a shame tolimit local connectivity because of the cost of peering with the rest ofthe world but that’s the way prices are often set today. There are a number of ways to address this:One is to limit the capacity at the peering point so that you would payfor a percentage of the capacity. This is similar to what we do now so maybe it’s better to look at it in reverse – we continue the bandwidth pricing but don’t put a cap on local capacity. A better way is to assume that the aggregated buying capacity of the community is sufficiently large so that there is relatively abundantcapacity. This is fine except that the video bits may be problematic butif other applications work and video doesn’t work very well then there is an Akamai-like opportunity. In fact, since we often find 50:1 to 100:1 ratios today we are already deploying such solutions. If we take this path then we can increase the effective capacity bypeering with adjacent communities or using dark fiber to reduce the costof peering with other cooperating communities. At some point these communities become sufficiently large that we repeat the phenomenon ofthe local community switching over to paying for physical infrastructurerather than leasing bandwidth. This is what I consider the end game – investing in the physical infrastructure by hiring companies to manage and deploy it. We would already be there except for the presumption that bandwidth pricing is the way to go because we are presuming a telecom model of services rather than the Internet in which we do our own network and just need physical facilities.We can also step back and look at this from a societal point of view. I keep noting that applying the ARPU model to other infrastructure such as sewers doesn’t make sense – we know they don't have to be profit centersbecause their value is in terms of the external benefit to society.There’s a joke about which part of the body is much valuable, the brain makes it’s case but one organ simply goes on strike to prove its value.What does it mean that these cities are opting out of providingnetworking in favor of traditional telecoms? It means they can’t see thevalue in being able to use the infrastructure. One reason is that if they are locked into the charging for bandwidth model then the cost of using the transport is high. We might have abundant capacity but we’reafraid to let the meter run. This the need to meter all the usage makesit difficult to deploy applications without a rationale and justification for each one. Something simple like connecting parking meters requires a special purpose implementation tuned to the billingrelationship. You could use a cellular model but you wouldn’t want to dothat per meter unless you negotiate a special deal so you instead negotiation to be able to aggregate the bits with a local network and then use a shared cellular modem. If the cities understood the value both in what is possible and the reduced costs then instead of looking to offload the network to some company that thinks it can maintain sufficient control over the way we network so as to maintain a high bandwidth/service pricing model it would pay for maintaining the network. That’s actually what a city does when it does its own network anyway. It’s a viable business model for the companies that maintain the facilities. The real problem is that the cities don’t see the value inhaving the infrastructure and, in effect, turns the sewer system over toARPU pricing. And then accepts the synthetic scarcity as if it were an essential scarcity rather than just a way to maintain prices (as usual http://www.frankston.com/?name=AssuringScarcity). Perhaps this makes more sense than my general claim that (at a firstapproximation) there isn’t a business in transporting bits for hire andthat we should instead invest in the physical transport. It’s the end game if we follow this line of reasoning. There are many business possible so why are we trying to hard to charge the FCC and PTT’s with maintaining the one that is not viable in itself, and maximizes costs and minimizes are ability to get the full benefit of our fixed assets? -----Original Message----- From: David Farber [mailto:dave () farber net] Sent: Sunday, March 25, 2007 13:06 To: ip () v2 listbox comSubject: [IP] Re: There is no business case for broadband Interent [Wasre: Interconnect Scaling] Begin forwarded message: From: Max Tulyev <president () ukraine su> Date: March 25, 2007 10:53:13 AM EDT To: dave () farber net, dewayne () warpspeed com Subject: Re: [IP] There is no business case for broadband Interent [Was re: Interconnect Scaling] Dewayne,Sorry, but you are wrong about it. I'm doing my small business in Kiev,Ukraine. There is very popular some kind of broadband connection called"Home network". It is an huge (up to 20000 users in some companies) Ethernet-based network. The most close to this is FTTH technology. This is a stable and working business. Average client payment varying from $15 to $20 from company to company. There is no need in high rateinvestments - network can connect neighbor houses for its cost as few as$500 and lower. Costs for connecting a user inside building is less than$20, including the cost of switch port and a new cable to his flat. Now I'll explain why there is (and especially won't be) no real competition between homenets and others. Let's see:WiFi. It is oriented.mostly to mobile users, roaming and moving. That'swhy there is relatively high prices. You need to spend ~$200 to make s connected sphere 50m in diameter. Please, don't say me about "free" connection. There is nothing free in this world. Somebody should pay,and he will get his money back from you any case, sure. You will need tobuy radio frequency. You have to be aware of interference. The highestspeed of such connection is SHARED to all cell users 300mbit now and upto may be a gigabit in future. WiFi have well-known problems with packetlosses, jitter and other unstabilities. CaTV. Good if you have a legacy TV cable. It is more stable than WiFi, but still have problems with jitter (i.e. VoIP, games, etc durig highload). It have 2-3 STMs (i.e. less than 300-450mbit) SHARED to ALL (notan one cell) users and 1 SHARED STM (155mbit) from them. As there isfree and wider frequency resource, it might be up to 1-2 SHARED gbits inthe future.xDSL. Things are more happy. Now is up to 54Mbit PRIVATE connectivity tothe client, 1-2mbit PRIVATE from the client, but no more than 100Mbit inthe future due to frequency limitation and real cable length to the client. Nowdays it is not more than 7/1 PRIVATE mbit. So... Home ethernet network. It is 100Mbit BOTH SIDES (upstream and downstream) PRIVATE connectivity nowdays and up to 1Gbit PRIVATE connectivity to the client without changing cable infrastructure. More than enough? Take a look back, there is minimal comfort speed of the Internet access became ten times more every five years. This kind of network capable to high-quality videophones, IPTV, VoIP (as it is good controlled we implementing QoS well), video-on-demand, application leasing and so on. Also home network is a resource itself. People share information each other, play together, etc. So there is a very little number of people choose much cheaper xDSL or CaTV connection! And most of that, there IS money. Even for growing without investments at all. David Farber wrote:Begin forwarded message:From: Dewayne Hendricks <dewayne () warpspeed com>Date: December 15, 2006 11:36:14 PM JSTTo: Dewayne-Net Technology List <dewayne-net () warpspeed com>Subject: [Dewayne-Net] There is no business case for broadbandInterent[Was re: Interconnect Scaling]Reply-To: dewayne () warpspeed com[Note: This item comes from reader Bill St. Arnaud. DLH]From: "Bill St.Arnaud" <bill.st.arnaud () canarie ca>Date: December 14, 2006 5:06:58 AM PSTTo: <dewayne () warpspeed com>, "'Dewayne-Net Technology List'"<dewayne-net () warpspeed com>Subject: There is no business case for broadband Interent [Was re:Interconnect ScalingI would argue that it is not a question of capacity but whetherthere is asustainable business case - either in the core, or in the last mile.The reality is that broadband Internet is a brutal business withrazor thinmargins. With the advent of free Wifi Internet services, and now freebroadband services that we are seeing deployed in Europe - Inuk,SkyB, etc,you would be insane to enter the broadband business either as apublic goodcommunity open access facility, or a commercial enterprise. CableTV isfarmore lucrative (and frankly has higher demand) than broadbandInternet.As a side note I am in Sweden at the moment where there are manycommunityopen access fiber and wireless networks. It is interesting to seeTelia (the incumbent) is slowly taking over the management of many ofthese openaccess community networks. The electrical companies andcommunities aredesperate to unload them because they can't make a decent returnand scaredof their future prospects. We have seen the same problem in my hometownOttawa - where the city owned open access and fiber network is beingsold toBell Canada, for the same reasons that the city is losing money andare noteven able to cover their costs. Of course the incumbents aremaking theusual warm fuzzy statements about maintaining the open accesspolicy etcetc. We will see how long that lasts.In Canada it is the cablecos that are cleaning up and even giving thetelcosa hard time. But the telcos are discovering that buying themanagement ofopen access fiber/wifi networks is far cheaper than building your ownnetworkBill-----Original Message-----From: dewayne-net () warpspeed com [mailto:dewayne-net () warpspeed com] OnBehalf Of Dewayne HendricksSent: Wednesday, December 13, 2006 1:19 PMTo: Dewayne-Net Technology ListSubject: [Dewayne-Net] re: Interconnect Scaling[Note: This comment comes from reader Thomas Leavitt. DLH]From: Thomas Leavitt <thomas () thomasleavitt org>Date: December 12, 2006 8:06:47 PM PSTTo: dewayne () warpspeed comSubject: Re: [Dewayne-Net] Interconnect ScalingDewayne,Despite all the back and forth... it still isn't clear to me whetherthe Internet backbone, as most of the mainstream carriers / serviceproviders have structured their portions of it or are capable ofadapting, is capable of handling the vastly greater traffic load thatthe advent of P2P based HD video distribution networks willprecipitate. This seems to be a matter of:a) whether the interconnects between the various carrier / providernetwork backbones are diversified and distributed enough to preventchokepoints from developingb) whether the architectural optimizations of your average P2Pnetwork mean that vastly smaller amounts of traffic will actually runover the backbone and through any potential chokepoints than mightotherwise be the case... which also seems to be a dependent on whatpercentage of the traffic generated will consist of widely popular(and thus widely served) content versus specialized content that isless likely to be available from a nearby peer (within a networkbackbone boundary)I suppose that, given the rapid and relatively painless (at least,apparently so) emergence of Internet video over the past year and the(apparent) ease with which the additional traffic was accommodated,we shouldn't worry too much... does anyone have year over yearfigures on the total volume of Internet traffic and the percentageincrease year over year? I tried hunting this information down inGoogle, but wasn't successful....Regards,Thomas LeavittDewayne Hendricks wrote:[Note: This item comes from reader Mike O'Dell. DLH]From: mo () ccr org (Mike O'Dell)Date: December 12, 2006 9:23:38 AM PSTTo: dewayne () warpspeed comSubject: interconnect scalingthe mystical powers of "large peering points" still lives, i seefor the better part of a decade, large ISPs have used multiplepoint-to-point interconnects between their networks, situatedby the demands of network engineering, not the location of"peering points".the *only* magic in MAE-EAST (or MAE-WEST) was that once upona time, when the total traffic to exchange was a couple of T1sworth with 4-5 other ISPs, there was an economy of scale inbuying those T1s as a cross-town DS3 (or later, metro ethernet.)everybody got their DS3 to the same place and a switch was putthere to connect the dots. once the traffic between pairs ofISPs got big enough, the economic optimization that was MAE-EASTand MAE-WEST fell apart.that's all they ever were - a policy-neutral economic optimization.and one that went deeply sub-optimal many, many years ago(at least for large ISPs - they still get used by smallerplayers whose traffic exchange needs can still take advantageof that particular economy of scale.)-mo-------------------------------------You are subscribed as maxtul () netassist kiev uaTo manage your subscription, go tohttp://v2.listbox.com/member/?listname=ipArchives at: http://www.interesting-people.org/archives/interesting-people/-- WBR, Max Tulyev (MT6561-RIPE, 2:463/253@FIDO) ------------------------------------------- Archives: http://v2.listbox.com/member/archive/247/@now Powered by Listbox: http://www.listbox.com
-- WBR, Max Tulyev (MT6561-RIPE, 2:463/253@FIDO) ------------------------------------------- Archives: http://v2.listbox.com/member/archive/247/@now Powered by Listbox: http://www.listbox.com
Current thread:
- Re: There is no business case for broadband Internet [Was re: Interconnect Scaling] David Farber (Apr 01)
- <Possible follow-ups>
- Re: There is no business case for broadband Internet [Was re: Interconnect Scaling] David Farber (Apr 15)