From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from bobcat.rjmcmahon.com (bobcat.rjmcmahon.com [45.33.58.123]) (using TLSv1.2 with cipher ADH-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id 9B66A3B2A4 for ; Tue, 10 Oct 2023 04:24:58 -0400 (EDT) Received: from [192.168.1.59] (c-69-181-111-171.hsd1.ca.comcast.net [69.181.111.171]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by bobcat.rjmcmahon.com (Postfix) with ESMTPSA id 780E61B258; Tue, 10 Oct 2023 01:24:57 -0700 (PDT) DKIM-Filter: OpenDKIM Filter v2.11.0 bobcat.rjmcmahon.com 780E61B258 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=rjmcmahon.com; s=bobcat; t=1696926297; bh=E86g+KubBXO12wKIVN0xM5LlEDZ07lb+FpYPyX+xl1c=; h=In-Reply-To:References:Subject:From:Date:To:CC:From; b=ULkAWAUnibips4UMoVWemQQEEmNm57b1/LHAG67Y8bAi2eopzKIxUmgDyvJHqKJOa kjjO7bc72Dlh98dC9Fb9JaEJTTfYzMaeM6VrGzoNs93e3LBYK0n7Tm3dJEkVWq1fur ysSalyosdxOjKqeiS5ekvdytOw4KOrxjUT1cczLk= In-Reply-To: References: <18b0c0fca5a.df21b356967361.3801960253537018542@phillywisper.net> <2EB085CD-44EB-4664-9436-6077A106151A@gmx.de> <8c3ecc6b-272a-414e-86b3-a56a029ec70c@rjmcmahon.com> <88139a6c8a4220851d25a9cfa1185159@rjmcmahon.com> <5e89d58588cc0509e61b0696a22d4b6c@rjmcmahon.com> X-Referenced-Uid: 0001142f567702d5 Thread-Topic: Re: [NNagain] The non-death of DSL X-Is-Generated-Message-Id: true X-Blue-Identity: !l=887&o=43&fo=54334&pl=791&po=0&qs=PREFIX&f=HTML&m=!%3AYzY1YmQ4MGQtZDViOC00ZGNkLTg5YTktNDhhMDUwMTEwOWJj%3ASU5CT1g%3D%3AMDAwMTE0MmY1Njc3MDJkNQ%3D%3D%3AANSWERED&p=784&q=SHOW User-Agent: Android MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----QVME9CIRYKUOEPRPZ7Y5JUFUSH3WFF" Content-Transfer-Encoding: 7bit From: Robert McMahon Date: Tue, 10 Oct 2023 01:24:51 -0700 To: Sebastian Moeller CC: Dave Taht via Nnagain Message-ID: Subject: Re: [NNagain] The non-death of DSL X-BeenThere: nnagain@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: =?utf-8?q?Network_Neutrality_is_back!_Let=C2=B4s_make_the_technical_aspects_heard_this_time!?= List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 10 Oct 2023 08:24:58 -0000 ------QVME9CIRYKUOEPRPZ7Y5JUFUSH3WFF Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=UTF-8 The Markeley Group in Boston charges $2500 to $5000 per month to lease a da= rk fiber strand into their colo=2E The labor to install is on the wanna be = isp=2E One has to provide the colo switches and pay Colo fees, peering fees= and transit fees=2E The only long lived parts are the fiber, patch panels = and building which are not sufficient by themselves to operate a modern ISP= =2E Switch tech is upgrading on silicon mfg cycles=2E Similar with optics t= hough the MTBF is less than silicon, hence pluggables=2E And that's just in= terconnections and ignores compute=2E https://www=2Emarkleygroup=2Ecom/ T= hen there are OSP pedestals and actives, optical amps e=2Eg=2E via doped fi= ber,=C2=A0 too=2E Operating an optical OSP requires a skilled labor force= =2E I don't see a muni funding model that works=2E Maybe in 20 or 30 years= but not today=2E I wish it were different=2E I think it is nice as an idea= l but the economics don't seem to work=2E If they did, we would see the mun= is everywhere already=2E It's been proposed for over two decades now and no= thing substantial has been built=2E Bob On Oct 9, 2023, 11:13 PM, at 11:1= 3 PM, Sebastian Moeller wrote: >Hi Bob=2E > >On 10 Octo= ber 2023 02:13:18 CEST, Robert McMahon > wrote: = >>Hi Sebastian, >> >>The NRE per chip starts at $100M=2E It's multiplr semi= conductors that >now define a networks and data centers capabilitied=2E A s= mall municipal >overbuilder is not a market maker=2E > >[SM] Sure, a small = outfit is essentially forced to use off the shelf >components, though they = might innovate a bit on the software side, like >libreqos=2E But for one of= the biggest current challenges, getting fiber >out to all residences/busin= esses is that really an issue? > > >> >>So yes, an overbuilder that can't f= und ASIC NRE needs to be intimately >aware of both market dynamics and the = state of engineering, of today, >tomorrow and the next 20-30 years as that'= s typically the life of the >municipal bonds=2E > >[SM] In a dark fiber mod= el, this will not matter too much, no? For a >lighted fiber approach 20-30 = years require 2-4 technology generations >which seems hard to predict=2E=2E= =2E then again the biggest cost is likely >the fiber access network, the ac= tive tech might by financable out of >the cash flow? > > >> >>Investors are= n't govt=2E bond holders and investor owned companies can >take more risk= =2E If low latency offerings don't increase ARPU, the >investors lose=2E If= it works, they win=2E Big difference=2E > >[SM] Building that fiber plant = seems like a pretty save bet to me, >allowing for longterm financing=2E Int= erestigly over here some >insurrances got into the FTTH build-out game, obv= iously considering it >a viable long term investment, though population den= sity is higher here >than in the US likely affecting cost and amortisation = periods=2E=2E=2E > >> >>=E2=81=A3Bob >> >>On Oct 9, 2023, 12:40 AM, at 12:4= 0 AM, Sebastian Moeller > wrote: >>>Hi Bob, >>> >>>> On = Oct 8, 2023, at 22:44, rjmcmahon >wrote: >>>> >= >>> Yeah, I get it=2E I think we're just too early for a structural >>>sepa= ration model in comm infrastructure=2E >>> >>>[SM] I see one reason why we = should not wait, and that is the >>>future-proofness of the eventually reac= hed FTTH-deployment=2E=2E=2E >>> >>> >>>> >>>> I think when we get to mix &= match DSP/optics and point to point >>>fiber in the OSPs, as done in data = centers, it may change=2E But today >>>it's PON at best which implies a com= munal decision process vs >>>individual one=2E >>> >>>[SM] There are IMHO t= wo components to the AON (Point to >>>MultiPoint/PtMP)/PON (Point to Point/= PtP) debate: >>>a) PONs are cheaper to operate, as they require less power = (on the >ISP >>>side) and space (depending on where the passive splitters a= re >located)=2E >>>b) Structural PONs with splitters out in the field (to r= ealize space >>>saving in the CO) are less flexible, one can always operate= a PtMP >>>plant as PON, but converting a structural PON to AON likely requ= ires >>>putting new fibers into the ground=2E >>> >>>The first part is some= thing I am not too concerned about, the coming >>>FTTH access network is go= ing to operate for decades, so I could not >>>care less about what active t= echnology is going to be used in the >next >>>decade (I assume that ISPs tr= yto keep the same tech operational for ~ >a >>>decade, but for PON that mig= ht be too pessimistic), the second part >is >>>different though=2E=2E=2E mi= cro-economics favor PtMP with splitters in the >>>field (lower up front cos= t* AND less potential for regulatory >>>intervention**) while the macro-eco= nomic perspective makes PtP more >>>attractive (offering more flexibility o= ver the expected life time of >>>multiple decades)=2E >>> >>> >>> >>>*) One= big item, the cost for actually deploying the fiber to is not >>>all that = sensitive, if you put fiber into the ground the traditional >>>way, typical= ly the cost of the earth works dominates over e=2Eg=2E the >cost >>>of the = individual fibers (not that this would stop bean-counter types >>>to still = minimize the number of fiber cables=2E=2E=2E)=2E >>> >>>**) With PtP the po= tential exists that a regulator (likely not the >FCC) >>>could force an ISP= to offer dark-fibers to end customers at wholesale >>>prices, with PtMP th= e ISP having build out likely will stay in >control >>>of the active tech i= n each segment (might be forced to offer >bitstream >>>access***) so potent= ial competitors will not be able to offer >>>better/faster technology on th= e shared fiber=2E That is some of the >PONs >>>are backward compatible and = in theory on the same PON tree one ISP >>>might be operating GPON while ano= ther ISP might theoreticallu offer >>>XGS-PON on the same segment, but I th= ink this is a rather theoretical >>>construct unlikely to happen quantitati= vely=2E=2E=2E >>> >>>***) Not only control of the tech, but offering bitstr= eam access >likely >>>means a larger wholesale price as well=2E >>> >>> >>>= > Communal actions, as seen in both LUS and Glasgow, can take decades >>>an= d once done, are slow to change=2E >>> >>> [SM] LUS already offer symmetric= 10G links=2E=2E=2E they do not seem to be >>>lagging behind, the main crit= icism seems to be that they are somewhat >>>more expensive than the big ISP= s, which is not all that surprising >>>given that they will not be able to = leverage scale effects all that >>>much simply by being small=2E=2E=2E Also= a small ISP likely can not afford >a >>>price war with a much larger compa= ny (that can afford to serve below >>>cost in areas it competes with smalle= r ISPs in an attempt to drive >>>those smaller ones out of the market, afte= r which prices likely >>>increase again)=2E >>> >>> >>>> The decision proce= ss time vs tech timelines exacerbate this=2E >Somebody >>>has to predict th= e future - great for investors & speculators, not so >>>for regulators look= ing backwards=2E >>> >>> [SM] I am not convinced that investors/speculators= actually do a >much >>>better job predicting the future, just look at how = much VC is wadted >on >>>hare-brained schemes like NFTs/crypto currencies a= nd the like? >>> >>>> Also, engineering & market cadence matching is critic= al and neither >>>LUS nor Glasgow solved that=2E >>> >>> [SM] But do they n= eed to solve that? Would it not be enough to >simply >>>keep offering somet= hing that in their service area is considered good >>>enough by the custome= rs? Whether they do or do not, I can not tell=2E >>> >>> >>>Regards >>> Seb= astian >>> >>> >>>> >>>> Bob >>>>> Hi Bob, >>>>>> On Oct 8, 2023, at 21:27,= rjmcmahon >>>wrote: >>>>>> Hi Sebastian, >>>>>= > Here's a good link on Glasgow, KY likely the first U=2ES=2E muni >>>netwo= rk started around 1994=2E It looks like a one and done type >>>investment= =2E Their offering was competitive for maybe a decade and now >>>seems to h= ave fallen behind for the last few decades=2E >>>>>> https://www=2Eglasgowe= pb=2Ecom/internet-packages/ >>>>>> https://communitynets=2Eorg/content/birt= h-community-broadband-video >>>>> [SM] Looks like they are using DOCSIS an= d are just about to go >>>fiber; >>>>> not totally unexpected, it takes awh= ile to amortize the cost of >say >>>a >>>>> CMTS to go DOCSIS and only afte= r that period you make some profit, >>>so >>>>> many ISPs will be tempted t= o operate the active gear a bit longer >>>>> longer after break even, as wi= th new active gear revenue will >likely >>>>> not generate surplus=2E The c= hallenge is to decide when to >upgrade=2E=2E=2E >>>>> My preferred model ho= wever is not necessarily having a communal >ISP >>>>> that sells internet a= ccess services (I am not against that), but >>>have >>>>> a communal built-= out of the access network and centralize the >lines >>>>> (preferably fiber= ) in a few large enough local IXs, so internet >>>access >>>>> providers on= ly need to bring their head-ends and upstream links to >>>>> those location= s to be able to offer services=2E In the beginning it >>>>> makes probably = sense to also offer some sort of GPON/XGSPON bit >>>stream >>>>> access to = reduce the up-front cost for ISPs that expect to serve >>>only >>>>> a smal= l portion of customers in such an IX, but that is pure >>>>> speculation=2E= =2E=2E=2E The real idea is to keep those things that will >>>result >>>>> i= n a natural monopoly to form in communal hands (that already >manage >>>>> = other such monopoly infrastructure anyway) and then try to use the >>>>> fa= ct that there is no local 800lb Gorilla ISP owning most lines to >>>try >>>= >> to create a larger pool of competing ISPs to light up the fiber >>>>> in= frastructure=2E=2E=2E That is I am fine with a market solution, if we >>>ca= n >>>>> assure the market to be big enough to actually deliver on its >>>>>= promises=2E >>>>>> LUS is similar if this article is to be believed=2E >>>= https://thecurrentla=2Ecom/2023/column-lus-fiber-has-lost-its-edge/ >>>>> = [SM] The article notices that comparing things is hard=2E=2E=2E as the >>>>= > offers differ considerably from what alternate ISPs offer (e=2Eg=2E >LUS = >>>>> offers symmetric capacity for down- and upload) and the number >>>>> = compared seems to be the advertised price, which IIRC in the US is >>>>> co= nsiderably smaller than what one happens to actually pay month >per >>>>> m= onth due to additional fees and stuff=2E=2E=2E (in Germany prices for >>>>>= end-customers typically are "all inclusive prices", the amount of >>>>> VA= T/tax is shown singled out in the receipts, but the number we >>>>> operate= on is typically the final price, but then we have almost >no >>>>> local t= axes that could apply)=2E >>>>>> The LUS NN site says there is no congestio= n on their fiber (GPON) >>>so they don't need AQM or other congestion mgmt = mechanisms which I >find >>>suspect=2E https://www=2Elusfiber=2Ecom/net-neu= trality >>>>> [SM] Actually intriguing, would I live in their area I would= try >>>them >>>>> out, then I could report on the details here :) >>>>> Br= owsing their documentation I am not a big fan of their volume >>>limits >>>= >> though, I consider these to be absurd measures of >control=2E=2E=2E=2E(a= bsurd >>>>> in that they are too loosely coupled with the relevant measure = for >>>the >>>>> actual cost)=2E >>>>>> This may demonstrate that technolog= y & new requirements are >moving >>>too quickly for municipal approaches=2E= >>>>> [SM] That might well be true=2E I have no insight any more on how >= >>this >>>>> affects commercial ISPs in the US either (I only tried two any= way >>>>> sonic and charter) >>>>>> Bob >>>>>>> Hi Sebastian, >>>>>>> The U= =2ES=2E of late isn't very good with regulatory that motivates >>>>>>> inve= stment into essential comm infrastructure=2E It seems to go >the >>>>>>> ot= her way, regulatory triggers under investment, a tragedy of >the >>>>>>> co= mmons=2E >>>>>>> The RBOCs eventually did overbuild=2E They used wireless a= nd went >to >>>>>>> contract carriage, and special access rate regulation h= as been >>>>>>> removed=2E The cable cos did HFC and have always been contr= act >>>carriage=2E >>>>>>> And they are upgrading today=2E >>>>>>> The tech= companies providing content & services are doing fine >too >>>and >>>>>>> = have enough power to work things out with the ISPs directly=2E >>>>>>> The = undeserved areas do need support=2E The BEAD monies may help=2E >I >>>>>>> = think these areas shouldn't be relegated to DSL=2E >>>>>>> Bob >>>>>>> On O= ct 8, 2023, at 2:38 AM, Sebastian Moeller >>>wrote: >>>= >>>>> Hi Bob, >>>>>>>> On 8 October 2023 00:13:07 CEST, rjmcmahon via Nnaga= in >>>>>>>> wrote: >>>>>>>>> Everybody = abandoned my local loop=2E Twisted pair from multiple >>>>>>>>> decades ago= into antiquated, windowless COs with punch blocks, >>>>>>>>> with no space= nor latency advantage for colocated content & >>>>>>>>> compute, seems to = have killed it off=2E >>>>>>>> [SM] Indeed, throughput for DSL is inversely= proportional to >loop >>>>>>>> length, so providing 'acceptable' capacity = requires >sufficiently >>>>>>>> short wire runs from DSLAM to CPE, and that= in turn means >moving >>>>>>>> DSLAMs closer to the end users=2E=2E=2E whi= ch in a densely populated >>>area >>>>>>>> works well, but in a less densel= y populated area becomes costly >>>>>>>> fast=2E And doing so will only mak= e sense if you get enough >>>customers >>>>>>>> on such an 'outdoor DSLAM' = so might work for the first to built >>>out, >>>>>>>> but becomes prohibiti= vely unattractive for other ISP later=2E >>>However >>>>>>>> terminating th= e loops in the field clears up lots of spaces in >>>the >>>>>>>> COs=2E=2E= =2E not that anybody over here moved much compute into >>>these=2E=2E=2E >>= >>>>>> (there exist too many COs to make that an attractive >proposition >>= >in >>>>>>>> spite of all the hype about moving compute to the edge)=2E As = is >a >>>few >>>>>>>> well connected data centers for compute seem to work = well >>>enough=2E=2E=2E >>>>>>>> I suspect in some towns one can buy out th= e local loop copper >>>with >>>>>>>> just a promise of maintenance=2E >>>>>= >>> [SM] A clear sign of regulatory failure to me, maintenance of >the >>>>= >>>> copper plant inherited from Bell should never have been left to >>>the= >>>>>>>> ISPs to decide about=2E=2E=2E >>>>>>>> The whole CLEC open the lo= op to competitive access seems to >have >>>>>>>> failed per costs, antiquat= ed technology, limited colocation, an >>>>>>>> outdated waveguide (otherwis= e things like CDDI would have won >>>over >>>>>>>> Cat 5), and market reaso= ns=2E The early ISPs didn't collocate, >they >>>>>>>> bought T1s and E1s an= d connected the TDM to statistical >>>multiplexingThe FCC has stated that = =E2=80=9Crigging or slanting the news is >a >>>most heinous act against the= public interest=2E=E2=80=9D >>>>>>>> - no major investment there either=2E= >>>>>>>>> The RBOCs, SBC (now AT&T) & and VZ went to contract carriage >an= d >>>>>>>>> wireless largely because of the burdens of title II per >>>regu= lators >>>>>>>>> not being able to create an investment into the OSPs=2E Th= e 2000 >>>>>>>>> blow up was kinda real=2E >>>>>>>> [SM] Again, I see no fa= ult in title 2 here, but in letting ISPs >>>of >>>>>>>> the hook on maintai= ning their copper plant or replace it with >>>>>>>> FTTH=2E=2E=2E >>>>>>>>>= She starts out by complaining about trying to place her WiFi >in >>>>>>>>>= the right place=2E That's like trying to share a flashlight=2E She >>>has = >>>>>>>>> access to the FCC technology group full of capable engineers=2E >= >>They >>>>>>>>> should have told her to install some structured wire, plac= e >more >>>>>>>>> APs, set the carrier and turn down the power=2E >>>>>>>> = [SM] I rather read this more as an attempt to built a report >with >>>>>>>>= the audience over a shared experience and less as a problem >>>report >>>>= >>>> ;) >>>>>>>> My wife works in the garden now using the garden AP SSID w= ith >no >>>>>>>> issues=2E My daughter got her own carrier too per here Dad= >>>dedicating >>>>>>>> a front end module for her distance learning needs= =2E I think her >>>>>>>> story to justify title II regulation is a bit made= up=2E >>>>>>>> [SM] Hmm, while covid19 lockdown wasn't the strongest examp= le, >I >>>>>>>> agree, I see no good argument for keeping essential >>>infr= astructure >>>>>>>> like internet access in private hands without appropria= te >>>oversight=2E >>>>>>>> Especially given the numbers for braodband choi= ce for >customers, >>>>>>>> clearly the market is not going to solve the is= sues at hand=2E >>>>>>>>> Also, communications have been essential back bef= ore the rural >>>>>>>>> free delivery of mail in 1896=2E Nothing new here o= ther than >>>>>>>>> hyperbole to justify a 5 member commission acting as th= e >single >>>>>>>>> federal regulator over 140M households and 33M business= es, >>>almost >>>>>>>>> none of which have any idea about the complexities = of the >>>>>>>>> internet=2E >>>>>>>> [SM] But the access network is quite = different than the >>>internet's >>>>>>>> core, so not being experts on the= core seems acceptable, no? >And >>>>>>>> even 5 members is clearly superio= r to no oversight at all? >>>>>>>> I'm not buying it and don't want to hand= the keys to the FCC >who >>>>>>>> couldn't protect journalism nor privacy= =2E Maybe start there, >>>looking >>>>>>>> at what they didn't do versus bl= aming contract carriage for a >>>>>>>> distraction? >>>>>>>> [SM] I can spe= ak to the FCC as regulatory agency, but over here >>>IMHO >>>>>>>> the nati= onal regulatory agency does a decent job arbitrating >>>between >>>>>>>> th= e interests of both sides=2E >>>>>>> >>>https://about=2Eusps=2Ecom/who/prof= ile/history/rural-free-delivery=2Ehtm#:~:text=3DOn%20October%201%2C%201896%= 2C%20rural,were%20operating%20in%2029%20states=2E >>>>>>>> Bob >>>>>>>> My = understanding, though I am not 100% certain, is that the >baby >>>>>>>> bel= ls >>>>>>>> lobbied to have the CLEC equal access provisions >revoked/gutte= d=2E >>>>>>>> Before this, the telephone companies were required to provide= >>>access >>>>>>>> to the "last mile" of the copper lines and the switches= at >>>wholesale >>>>>>>> costs=2E Once the equal access provisions were re= moved, the >>>telephone >>>>>>>> companies started charging the small phone= and DSL providers >>>close >>>>>>>> to >>>>>>>> the retail price for acces= s=2E The CLEC DSL providers could not >>>stay >>>>>>>> in >>>>>>>> business= when they charged a customer $35 / month for Internet >>>>>>>> service >>>= >>>>> while the telephone company charged the DSL ISP $35 / month for >>>>>= >>> access=2E >>>>>>>> ---- On Sat, 07 Oct 2023 17:22:10 -0400 Dave Taht vi= a Nnagain >>>wrote >>>>>>>> --- >>>>>>>> I have a lot to unpack from this: = >>>>>>>> https://docs=2Efcc=2Egov/public/attachments/DOC-397257A1=2Epdf >>>= >>>>> the first two on my mind from 2005 are: "FCC adopted its first >>>ope= n >>>>>>>> internet policy" and "Competitiveness" As best as I recall, >(an= d >>>>>>>> please correct me), this led essentially to the departure of >al= l >>>the >>>>>>>> 3rd party DSL providers from the field=2E I had found som= ething >>>>>>>> referencing this interpretation that I cannot find right no= w, >but >>>I >>>>>>>> do >>>>>>>> clearly remember all the DSL services you= could buy from in the >>>>>>>> early >>>>>>>> 00s, and how few you can buy= from now=2E Obviously there are many >>>>>>>> other >>>>>>>> possible root= causes=2E >>>>>>>> DSL continued to get better and evolve, but it definate= ly >suffers >>>>>>>> from >>>>>>>> many reports of degraded copper quality,= but does an estimate >>>exist >>>>>>>> for how much working DSL is left? >= >>>>>>> Q0) How much DSL is in the EU? >>>>>>>> Q1) How much DSL is left in= the USA? >>>>>>>> Q2) What form is it? (VDSL, etc?) >>>>>>>> Did competiti= on in DSL vanish because of or not of an FCC >related >>>>>>>> order? >>>>>= >>> -- >>>>>>>> Oct 30: >>>>>>>> >>>https://netdevconf=2Einfo/0x17/news/the= -maestro-and-the-music-bof=2Ehtml >>>>>>>> Dave T=C3=A4ht CSO, LibreQos >>>= >>>>> ------------------------- >>>>>>>> Nnagain mailing list >>>>>>>> Nnag= ain@lists=2Ebufferbloat=2Enet >>>>>>>> https://lists=2Ebufferbloat=2Enet/li= stinfo/nnagain >>>>>>>> ------------------------- >>>>>>>> Nnagain mailing = list >>>>>>>> Nnagain@lists=2Ebufferbloat=2Enet >>>>>>>> https://lists=2Ebu= fferbloat=2Enet/listinfo/nnagain >>>>>>> ------------------------- >>>>>>> = Nnagain mailing list >>>>>>> Nnagain@lists=2Ebufferbloat=2Enet >>>>>>> http= s://lists=2Ebufferbloat=2Enet/listinfo/nnagain >>>>>>> ____________________= ___________________________ >>>>>>> Nnagain mailing list >>>>>>> Nnagain@li= sts=2Ebufferbloat=2Enet >>>>>>> https://lists=2Ebufferbloat=2Enet/listinfo/= nnagain > >-- >Sent from my Android device with K-9 Mail=2E Please excuse = my brevity=2E ------QVME9CIRYKUOEPRPZ7Y5JUFUSH3WFF Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: quoted-printable
The Markeley= Group in Boston charges $2500 to $5000 per month to lease a dark fiber str= and into their colo=2E The labor to install is on the wanna be isp=2E One h= as to provide the colo switches and pay Colo fees, peering fees and transit= fees=2E The only long lived parts are the fiber, patch panels and building= which are not sufficient by themselves to operate a modern ISP=2E Switch t= ech is upgrading on silicon mfg cycles=2E Similar with optics though the MT= BF is less than silicon, hence pluggables=2E And that's just interconnectio= ns and ignores compute=2E

https://www=2Emarkleygroup=2Ecom/

Then there are OSP pedestals and actives, optical am= ps e=2Eg=2E via doped fiber,=C2=A0 too=2E Operating an optical OSP requires= a skilled labor force=2E

I don't see a mun= i funding model that works=2E Maybe in 20 or 30 years but not today=2E I wi= sh it were different=2E I think it is nice as an ideal but the economics do= n't seem to work=2E If they did, we would see the munis everywhere already= =2E It's been proposed for over two decades now and nothing substantial has= been built=2E

Bob
On Oct 9, 2023, at 11:13 PM, Sebastian Moeller <moeller0@gmx=2Ede> wrote: 
Hi Bob=2E

On 10 October 2023 02:13:18 CEST, Robert McMahon <r=
jmcmahon@rjmcmahon=2Ecom> wrote:
Hi Sebastian,

The NRE per chip starts at $100M=2E It's =
multiplr semiconductors that now define a networks and data centers capabil=
itied=2E A small municipal overbuilder is not a market maker=2E

[SM] Sure, a small outfit is essentially forced to use off the she=
lf components, though they might innovate a bit on the software side, like =
libreqos=2E But for one of the biggest current challenges, getting fiber ou=
t to all residences/businesses is that really an issue?



So yes, an overbuilder that ca=
n't fund ASIC NRE needs to be intimately aware of both market dynamics and =
the state of engineering, of today, tomorrow and the next 20-30 years as th=
at's typically the life of the municipal bonds=2E

[SM] =
In a dark fiber model, this will not matter too much, no? For a lighted fib=
er approach 20-30 years require 2-4 technology generations which seems hard=
 to predict=2E=2E=2E then again the biggest cost is likely the fiber access=
 network, the active tech might by financable out of the cash flow?

=


Investors aren't g=
ovt=2E bond holders and investor owned companies can take more risk=2E If l=
ow latency offerings don't increase ARPU, the investors lose=2E If it works=
, they win=2E Big difference=2E

[SM] Building that fibe=
r plant seems like a pretty save bet to me, allowing for longterm financing=
=2E Interestigly over here some insurrances got into the FTTH build-out gam=
e, obviously considering it a viable long term investment, though populatio=
n density is higher here than in the US likely affecting cost and amortisat=
ion periods=2E=2E=2E

Hi Bob,

 On Oct 8, 2023, at 22:44, rjmcmahon <rjmcmahon@rjmcmahon=2Ecom&g=
t; wrote:

 Yeah, I get it=2E I think we're just too early for a stru=
ctural
separation model in comm infrastructure=2E

[S=
M] I see one reason why we should not wait, and that is the
future-proof=
ness of the eventually reached FTTH-deployment=2E=2E=2E



 I think when we get to mix &a=
mp; match DSP/optics and point to point
fiber in the OSPs, =
as done in data centers, it may change=2E But today
it's PON at best whi=
ch implies a communal decision process vs
individual one=2E

[SM] =
There are IMHO two components to the AON (Point to
MultiPoint/PtMP)/PON =
(Point to Point/PtP) debate:
a) PONs are cheaper to operate, as they req=
uire less power (on the ISP
side) and space (depending on where the pass=
ive splitters are located)=2E
b) Structural PONs with splitters out in t=
he field (to realize space
saving in the CO) are less flexible, one can =
always operate a PtMP
plant as PON, but converting a structural PON to A=
ON likely requires
putting new fibers into the ground=2E

The firs=
t part is something I am not too concerned about, the coming
FTTH access=
 network is going to operate for decades, so I could not
care less about=
 what active technology is going to be used in the next
decade (I assume=
 that ISPs tryto keep the same tech operational for ~ a
decade, but for =
PON that might be too pessimistic), the second part is
different though=
=2E=2E=2E micro-economics favor PtMP with splitters in the
field (lower =
up front cost* AND less potential for regulatory
intervention**) while t=
he macro-economic perspective makes PtP more
attractive (offering more f=
lexibility over the expected life time of
multiple decades)=2E


*) One big item, the cost for actually deploying the fiber to is not<=
br>all that sensitive, if you put fiber into the ground the traditional
=
way, typically the cost of the earth works dominates over e=2Eg=2E the cost=

of the individual fibers (not that this would stop bean-counter typesto still minimize the number of fiber cables=2E=2E=2E)=2E

**) With=
 PtP the potential exists that a regulator (likely not the FCC)
could fo=
rce an ISP to offer dark-fibers to end customers at wholesale
prices, wi=
th PtMP the ISP having build out likely will stay in control
of the acti=
ve tech in each segment (might be forced to offer bitstream
access***) s=
o potential competitors will not be able to offer
better/faster technolo=
gy on the shared fiber=2E That is some of the PONs
are backward compatib=
le and in theory on the same PON tree one ISP
might be operating GPON wh=
ile another ISP might theoreticallu offer
XGS-PON on the same segment, b=
ut I think this is a rather theoretical
construct unlikely to happen qua=
ntitatively=2E=2E=2E

***) Not only control of the tech, but offering=
 bitstream access likely
means a larger wholesale price as well=2E

 Communal actions, a=
s seen in both LUS and Glasgow, can take decades
and once d=
one, are slow to change=2E

 [SM] LUS already offer symmetric 10G lin=
ks=2E=2E=2E they do not seem to be
lagging behind, the main criticism se=
ems to be that they are somewhat
more expensive than the big ISPs, which=
 is not all that surprising
given that they will not be able to leverage=
 scale effects all that
much simply by being small=2E=2E=2E Also a small=
 ISP likely can not afford a
price war with a much larger company (that =
can afford to serve below
cost in areas it competes with smaller ISPs in=
 an attempt to drive
those smaller ones out of the market, after which p=
rices likely
increase again)=2E


 The decision process time vs tech timelines exacerbate=
 this=2E Somebody
has to predict the future - great for inv=
estors & speculators, not so
for regulators looking backwards=2E
=

 [SM] I am not convinced that investors/speculators actually do a much<=
br>better job predicting the future, just look at how much VC is wadted on<=
br>hare-brained schemes like NFTs/crypto currencies and the like?

 Also, engineering & mar=
ket cadence matching is critical and neither
LUS nor Glasgo=
w solved that=2E

 [SM] But do they need to solve that? Would it not =
be enough to simply
keep offering something that in their service area i=
s considered good
enough by the customers? Whether they do or do not, I =
can not tell=2E


Regards
 Sebastian



 Bob
 Hi Bob,
 On Oct 8, 2023, at 21:27, rjmcmahon <rjmcmahon@rjmcmahon=2Ecom>=
;
wrote:
 Hi Sebastian,
 Here'=
s a good link on Glasgow, KY likely the first U=2ES=2E muni
network started around 1994=2E It looks like a o=
ne and done type
investment=2E Their offering was competitive for maybe =
a decade and now
seems to have fallen behind for the last few decades=2E=

 https://www=
=2Eglasgowepb=2Ecom/internet-packages/
 https://communitynets=
=2Eorg/content/birth-community-broadband-video
  [SM] L=
ooks like they are using DOCSIS and are just about to go
fiber;
 not totally unexpected, it tak=
es awhile to amortize the cost of say
a
 CMTS to go DOCSIS and only after that period you make =
some profit,
so
 man=
y ISPs will be tempted to operate the active gear a bit longer
 longer a=
fter break even, as with new active gear revenue will likely
 not genera=
te surplus=2E The challenge is to decide when to upgrade=2E=2E=2E
 My pr=
eferred model however is not necessarily having a communal ISP
 that sel=
ls internet access services (I am not against that), but
have
 a communal built-out of the acce=
ss network and centralize the lines
 (preferably fiber) in a few large e=
nough local IXs, so internet
access
 providers only need to bring their head-ends and upstream =
links to
 those locations to be able to offer services=2E In the beginni=
ng it
 makes probably sense to also offer some sort of GPON/XGSPON bit
stream
 access to red=
uce the up-front cost for ISPs that expect to serve
only
 a small portion of customers in such =
an IX, but that is pure
 speculation=2E=2E=2E=2E The real idea is to kee=
p those things that will
result
 in a natural monopoly to form in communal hands (that already =
manage
 other such monopoly infrastructure anyway) and then try to use t=
he
 fact that there is no local 800lb Gorilla ISP owning most lines to
try
 to create a la=
rger pool of competing ISPs to light up the fiber
 infrastructure=2E=2E=
=2E That is I am fine with a market solution, if we
can
 assure the market to be big enough to =
actually deliver on its
 promises=2E
 LUS is similar if this article is to be believed=2E
https://thecurrentla=2Ecom/2023/co=
lumn-lus-fiber-has-lost-its-edge/
  [SM] The =
article notices that comparing things is hard=2E=2E=2E as the
 offers di=
ffer considerably from what alternate ISPs offer (e=2Eg=2E LUS
 offers s=
ymmetric capacity for down- and upload) and the number
 compared seems t=
o be the advertised price, which IIRC in the US is
 considerably smaller=
 than what one happens to actually pay month per
 month due to additiona=
l fees and stuff=2E=2E=2E (in Germany prices for
 end-customers typicall=
y are "all inclusive prices", the amount of
 VAT/tax is shown singled ou=
t in the receipts, but the number we
 operate on is typically the final =
price, but then we have almost no
 local taxes that could apply)=2E
<=
blockquote class=3D"gmail_quote" style=3D"margin: 0pt 0pt 1ex 0=2E8ex; bord=
er-left: 1px solid #e9b96e; padding-left: 1ex;"> The LUS NN site says there=
 is no congestion on their fiber (GPON)
so they don't need AQM or other congestion mgmt mechanisms which I f=
ind
suspect=2E h=
ttps://www=2Elusfiber=2Ecom/net-neutrality
  =
[SM] Actually intriguing, would I live in their area I would try
them
 out, then I could report=
 on the details here :)
 Browsing their documentation I am not a big fan=
 of their volume
limits
 though, I consider these to be absurd measures of control=2E=2E=2E=
=2E(absurd
 in that they are too loosely coupled with the relevant measu=
re for
the
 actual c=
ost)=2E
 This may demon=
strate that technology & new requirements are moving
too quickly for municipal approaches=2E
  [SM] That might well be true=2E I have no insight any more=
 on how
this
 affect=
s commercial ISPs in the US either (I only tried two anyway
 sonic and c=
harter)
 Bob
 Hi Sebastian,
 The U=2ES=2E of l=
ate isn't very good with regulatory that motivates
 investment into esse=
ntial comm infrastructure=2E It seems to go the
 other way, regulatory t=
riggers under investment, a tragedy of the
 commons=2E
 The RBOCs eve=
ntually did overbuild=2E They used wireless and went to
 contract carria=
ge, and special access rate regulation has been
 removed=2E The cable co=
s did HFC and have always been contract
carriage=2E
 And they are upgrading today=2E
 The tech companies providing co=
ntent & services are doing fine too
and
 have=
 enough power to work things out with the ISPs directly=2E
 The undeserv=
ed areas do need support=2E The BEAD monies may help=2E I
 think these a=
reas shouldn't be relegated to DSL=2E
 Bob
 On Oct 8, 2023, at 2:38 A=
M, Sebastian Moeller <moeller0@gmx=2Ede>
wrote:
 Hi Bob,
 On 8 Octo=
ber 2023 00:13:07 CEST, rjmcmahon via Nnagain
 <nnagain@lists=2Ebuffe=
rbloat=2Enet> wrote:
 E=
verybody abandoned my local loop=2E Twisted pair from multiple
 decades =
ago into antiquated, windowless COs with punch blocks,
 with no space no=
r latency advantage for colocated content &
 compute, seems to have =
killed it off=2E
 [SM] Indeed, throughput for DSL is invers=
ely proportional to loop
 length, so providing 'acceptable' capacity req=
uires sufficiently
 short wire runs from DSLAM to CPE, and that in turn =
means moving
 DSLAMs closer to the end users=2E=2E=2E which in a densely=
 populated
area
 works well, but in a less densely populated a=
rea becomes costly
 fast=2E And doing so will only make sense if you get=
 enough
customers
 on such an 'outdoor DSLAM' so might work for =
the first to built
<=
/blockquote>out,
 but becomes prohibitively unattractive f=
or other ISP later=2E
However
=

 terminating the loops in the field=
 clears up lots of spaces in
the
 COs=2E=2E=2E not that anybody o=
ver here moved much compute into
=

these=2E=2E=2E
 (there exist too=
 many COs to make that an attractive proposition
in
 spite of all=
 the hype about moving compute to the edge)=2E As is a
few
 well =
connected data centers for compute seem to work well
enough=2E=2E=2E
 I suspect in some towns one can buy out the local loop copper
with
 just a promise of maintenance=2E
 [SM] A clear sign of regulator=
y failure to me, maintenance of the
 copper plant inherited from Bell sh=
ould never have been left to
the
 ISPs to decide about=2E=2E=2E The whole CLEC open the loop to competitive access seems to have
 fai=
led per costs, antiquated technology, limited colocation, an
 outdated w=
aveguide (otherwise things like CDDI would have won
over
 Cat 5),=
 and market reasons=2E The early ISPs didn't collocate, they
 bought T1s=
 and E1s and connected the TDM to statistical
=

multiplexingThe FCC has stated that =
=E2=80=9Crigging or slanting the news is a
most heinous act against the =
public interest=2E=E2=80=9D
<=
blockquote class=3D"gmail_quote" style=3D"margin: 0pt 0pt 1ex 0=2E8ex; bord=
er-left: 1px solid #ccc; padding-left: 1ex;"> - no major investment there e=
ither=2E
 The RBOCs, SBC (=
now AT&T) & and VZ went to contract carriage and
 wireless large=
ly because of the burdens of title II per
regulators
 not being able to create a=
n investment into the OSPs=2E The 2000
 blow up was kinda real=2E
 [SM] Again, I see no fault in title 2 here, but in letting ISPs<=
br>
of
<=
blockquote class=3D"gmail_quote" style=3D"margin: 0pt 0pt 1ex 0=2E8ex; bord=
er-left: 1px solid #8ae234; padding-left: 1ex;">
 the hook on maintaining their copper plant or replace it with=

 FTTH=2E=2E=2E
 She st=
arts out by complaining about trying to place her WiFi in
 the right pla=
ce=2E That's like trying to share a flashlight=2E She
has
 access to the FCC tec=
hnology group full of capable engineers=2E
They
 should have told her to install=
 some structured wire, place more
 APs, set the carrier and turn down th=
e power=2E
 [SM] I rather read this more as an attempt to b=
uilt a report with
 the audience over a shared experience and less as a =
problem
report
 ;)
 My wife works in the garden now using =
the garden AP SSID with no
 issues=2E My daughter got her own carrier to=
o per here Dad
dedicating
 a front end module for her distance le=
arning needs=2E I think her
 story to justify title II regulation is a b=
it made up=2E
 [SM] Hmm, while covid19 lockdown wasn't the strongest exa=
mple, I
 agree, I see no good argument for keeping essential
infrastructure
=

 like internet access in private hands without appropriate
oversight=
=2E
 Especially given the numbers for braodband choice f=
or customers,
 clearly the market is not going to solve the issues at ha=
nd=2E
 Also, communicati=
ons have been essential back before the rural
 free delivery of mail in =
1896=2E Nothing new here other than
 hyperbole to justify a 5 member com=
mission acting as the single
 federal regulator over 140M households and=
 33M businesses,
almost
 none of which have any idea about the complexities of =
the
 internet=2E
 [SM] But the access network is quite d=
ifferent than the
internet's
<=
blockquote class=3D"gmail_quote" style=3D"margin: 0pt 0pt 1ex 0=2E8ex; bord=
er-left: 1px solid #fcaf3e; padding-left: 1ex;">
 core, so not being experts on the c=
ore seems acceptable, no? And
 even 5 members is clearly superior to no =
oversight at all?
 I'm not buying it and don't want to hand the keys to =
the FCC who
 couldn't protect journalism nor privacy=2E Maybe start ther=
e,
look=
ing
 at what they didn't do versus blaming contract carr=
iage for a
 distraction?
 [SM] I can speak to the FCC as regulatory a=
gency, but over here
IMHO
 the national regulatory agency does a =
decent job arbitrating
between
 the interests of both sides=2E

https://about=2Eusps=2Ecom/who/profile/history/rural-free-del=
ivery=2Ehtm#:~:text=3DOn%20October%201%2C%201896%2C%20rural,were%20operatin=
g%20in%2029%20states=2E
<=
blockquote class=3D"gmail_quote" style=3D"margin: 0pt 0pt 1ex 0=2E8ex; bord=
er-left: 1px solid #ccc; padding-left: 1ex;"> Bob
 My understanding, tho=
ugh I am not 100% certain, is that the baby
 bells
 lobbied to have t=
he CLEC equal access provisions revoked/gutted=2E
 Before this, the tele=
phone companies were required to provide
access
 to the "last mil=
e" of the copper lines and the switches at
wholesale
 costs=2E On=
ce the equal access provisions were removed, the
telephone
 compa=
nies started charging the small phone and DSL providers
close
 to=

 the retail price for access=2E The CLEC DSL providers could not
stay
 in
 business when they charged a customer $35 / month for Inte=
rnet
 service
 while the telephone company charged the DSL ISP $35 / =
month for
 access=2E
 ---- On Sat, 07 Oct 2023 17:22:10 -0400 Dave Ta=
ht via Nnagain
wrote
 ---
 I have a lot to unpack from this: https://docs=2Efcc=2Egov/public/attachments/DOC-397257A1=2Epdf
 t=
he first two on my mind from 2005 are: "FCC adopted its first
open
 ple=
ase correct me), this led essentially to the departure of all
the
 for how much working =
DSL is left?
 Q0) How much DSL is in the EU?
 Q1) How much DSL is lef=
t in the USA?
 Q2) What form is it? (VDSL, etc?)
 Did competition in =
DSL vanish because of or not of an FCC related
 order?
 --
 Oct 30=
:

https://netdevconf=2Einfo/0x17/news/the-maestro-and-the-music-bof=2Eht=
ml
 Dave T=C3=A4ht CSO, LibreQos
 ------------=
-------------
 Nnagain mailing list
 Nnagain@lists=2Ebufferbloat=2Ene=
t
 https:=
//lists=2Ebufferbloat=2Enet/listinfo/nnagain
 ----------------------=
---
 Nnagain mailing list
 Nnagain@lists=2Ebufferbloat=2Enet
 https://lists=2E=
bufferbloat=2Enet/listinfo/nnagain
 -------------------=
------
 Nnagain mailing list
 Nnagain@lists=2Ebufferbloat=2Enet
 <=
a href=3D"https://lists=2Ebufferbloat=2Enet/listinfo/nnagain">https://lists=
=2Ebufferbloat=2Enet/listinfo/nnagain

 Nnagain mailing list<=
br> Nnagain@lists=2Ebufferbloat=2Enet
 https://lists=2Ebufferbloat=2Enet/listinfo/nn=
again
------QVME9CIRYKUOEPRPZ7Y5JUFUSH3WFF--