January 2008 - NANOG - lists.nanog.org

Re: An Attempt at Economically Rational Pricing: Time Warner Trial
by odlyzko＠dtc.umn.edu 22 Jan '08

22 Jan '08

Such caps, if they are high enough, may be a reasonable compromise. As Mark Newton wrote a few days ago, about Australia, The more sensible end of town pays about $80 per month for about 40 Gbytes of quota, give or take, depending on the ISP. After that they get shaped to 64 kbps unless they want to pay more for more quota. Bytecounts are retrieved via SNMP (for business customers) or Radius (for DSL, dial, ISDN, etc). When transit is costing $250 per megabit per month, there aren't many other options. Given Australia's level of Internet traffic (see http://www.dtc.umn.edu/mints/) it seems that only a tiny fraction of the users will hit the 40 Gbytes of quota. But if your transit costs $10 per megabit per month, other factors may dominate. I have a discussion of these issues in the paper "Internet pricing and the history of communications," published in Computer Networks 36 (2001), pp. 493-517, available at http://www.dtc.umn.edu/~odlyzko/doc/history.communications1b.pdf Some of these issues are also dealt in the more recent paper with David Levinson, "Too expensive to meter: The influence of transaction costs in transportation and communication," Phil. Trans. Royal Soc. A, to appear, http://www.dtc.umn.edu/~odlyzko/doc/metering-expensive.pdf Overall, telecom policy makers, both inside service providers and in regulatory bodies, have been fixated on a particular economic model that denigrates flat rate plans. Now I am not a flat rate bigot, and understand their limitations. But it seems imperative to appreciate that there are several other factors that matter, discussed in the papers mentioned above. One is that people are willing to pay more for flat rates. Second is that flat rates stimulate usage, something that I claim telcos should be striging to do, as transmission capacity is growing. But few people appear willing to learn that lesson. Andrew > On Sun Jan 20, Matthew Moyle-Croft wrote: Simon Leinen wrote: > While I think this is basically a sound approach, I'm skeptical that > *slightly* lowering prices will be sufficient to convert 80% of the > user base from flat to unmetered pricing. Don't underestimate the > value that people put on not having to think about their consumption. > As long as the companies convince people that the "cap" is large enough to be essentially the same as unmetered then most people won't care and will take the savings. The other angle is to convince the 95% of customers that caps will actually deliver them a faster speed as the "evil 5%ers" won't be slowing them down by hogging the bandwidth. Having a cap and slowing down afterward (64kbps or 128kbps are typical) is what worked here in Oz. It also removes a whole lot of credit related issues. Consumers get a product where they know what they're getting - it's fast upto a point and then it slows down. -- Matthew Moyle-Croft - Internode/Agile - Networks Level 3, 132 Grenfell Street, Adelaide, SA 5000 Australia Email: mmc(a)internode.com.au Web: http://www.on.net Direct: +61-8-8228-2909 Mobile: +61-419-900-366 Reception: +61-8-8228-2999 Fax: +61-8-8235-6909 "The difficulty lies, not in the new ideas, but in escaping from the old ones" - John Maynard Keynes

16 32

Re: Lessons from the AU model
by Tom Vest 22 Jan '08

22 Jan '08

On Jan 21, 2008, at 8:20 PM, Roderick Beck wrote: > Tom, > > With all due respect, what are you trying to say in simple, clear > English? > > It is plain wrong to claim the DWDM can explain today's prices on > the Atlantic Is it? Sorry, how do you explain it then? An unanticipated/ undiscounted DWDM-driven windfall in salable supply (1), plus competition (2), plus financial distress (3), plus "innovative" transactions (4) are my prime suspects; the fact that every increment of private line capacity sold or otherwise transferred is fungible and can be further multiplexed just accelerates the (2-4) dynamic. I believe that the same thing -- (2) leading to price collapse, leading to (3) -- would have happened even if there been only two TA cables, assuming of course that (2) still applied -- maybe even if there were only one cable with enough upgradeable fiber pairs. Sure, "overinvestment" played some marginal role, but each new facilities platform added a multiple to the available supply, 1x, 2x... 8x... whereas DWDM advances that really kicked in only after gen1 entepreneurial cables were already planned if not in the water, added/ continues to add exponents (1x)^4, (2x)^6... (8x)^n? > and that DWDM allows providers to make sufficient ROI.. I didn't say that; I did say that I don't know how to define, much less calculate, "sufficient" ROI. Sufficient for what? When is "sufficient" ever sufficient, assuming you can extract more? If I can build one house, and rent it, manage it, and and maintain it at an overall return that will enable me to subsidize 100% of the purchase of another identical lot and identical materials and labor to build a second house, then maybe I'll decide that that's not really sufficient, e.g., because people demand bigger houses today, or maybe I think I could squeeze harder and get enough for three houses, or maybe four... Which one is "sufficient"? What if I've already got a lot of empty houses, but the market for housing won't deliver prices that satisfy my preferred sufficiency test? I have to believe that the financial authorities (public and private) will recognize these new, previously undiscounted developments, and find a way to integrate this new, previously unrecognized abundance into the broader economy -- e.g., by bankrolling future capacity builds (when there is real demand) even though their builders can't promise and won't deliver bubble-era returns. I have to believe it because I'm not cynical or paranoid enough to buy into any worldview that would lead to other assumptions. I might be proven wrong -- but if I do I'd have to start wondering about all sorted of things, e.g., the 200mpg combustion engine, etc... TV > About 70 percent of the operating costs of a TransAtlantic cable > are undersea maintenance and undersea maintenance costs are rising > (think fuel costs). > You can spread those fixed costs over more waves as DWDM improves, > but it is not enough in general. > I was comparing initial build/launch capex to the incremental capex required for upgrades (granted these costs vary by platform/ technology generation). I'm not surprised that a large/rising share of opex is O&M. The fuel component of O&M isn't incrementing in parallel with additional capacity sales, so why isn't expansion of the de facto sponsorship pool one way to offset the rising costs? TV > > Sent wirelessly via BlackBerry from T-Mobile. > > -----Original Message----- > From: Tom Vest <tvest(a)eyeconomics.com> > > Date: Mon, 21 Jan 2008 18:51:24 > To:Mark Newton <newton(a)internode.com.au> > Cc:roderickbeck@tmo.blackberry.net,owner-nanog@merit.edu,"Geoff > Huston" <gih(a)apnic.net>,"Matthew Moyle-Croft" > <mmc(a)internode.com.au>,"Randy Bush" <randy(a)psg.com>,"Andy Davidson" > <andy(a)nosignal.org>,"Andrew Odlyzko" > <odlyzko(a)dtc.umn.edu>,nanog(a)merit.edu > Subject: Re: Lessons from the AU model > > > > > On Jan 21, 2008, at 6:10 PM, Mark Newton wrote: > >> On 21/01/2008, at 10:49 PM, Tom Vest wrote: >> >>> In the absence of competition (and esp. in the presence of risk of >>> empowering competitive entrants), supply has no general/necessary >>> effect on prices at all. >>> So excess capacity of a product that is completely monopolized (or >>> priced by cartel fiat, ala OPEC or SC) is largely irrelevant. >> >> It goes a bit deeper than that when the monopoly can compound the >> problem my artificially constraining capacity by underspending on >> infrastructure (e.g., only lighting one pair on a multi-pair cable) >> >> So infrastructure spending can (and does) affect the price. > > Hi Mark, > > So you're saying that if a cable owner/monopolist simply lit another > fiber pair, that would cause them to reduce prices? > This mistakes a contingent symptom or tactic or mechanism -- i.e., > the intentionally misleading public explanation ("we're sold out") -- > with the real cause/strategy/motives behind "artificially" high > prices... > >> We get that every day in .au (Transmission on the monopoly route >> between Melbourne and Hobart costs 3 times more than transmission >> between Sydney and LA; and other potential cable operators have >> always known that the monopoly has an excess of supply hidden away >> somewhere which they can roll out at bargain basement prices if >> a competitor ever arrives in the market) >> >>> [ housing ] >>> Come to think of it, our sector has been struggling with its own >>> roughly similar terms-of-exchange crisis since about 2004-2005... >>> arguably driven by very similar prior circumstances as well... >>> worth investigating a bit further perhaps... >> >> I think the dominant factor that the American internet sector has >> been grappling with goes back further than that. It has its origins >> in the dot-com boom, when lots of people who didn't have any real >> money rolled out enormous infrastructure buildouts. When they >> inevitably went broke their infrastructure was bought at cents in >> the dollar, enabling the current generation of Internet companies >> to behave as if the infrastructure they're using was a lot cheaper >> than it really is. >> >> So hardly anyone has been selling below cost, but almost everyone >> has been selling below replacement cost. > > > This makes perfect sense for resources that are not subject to > "multiplexing effects" -- esp. unanticipated (undiscounted) > multiplexing effects. > Put it this way: how would you define (much less calculate) > "replacement cost" for an asset whose financing was predicated on a > useful of capacity of (x), but which, with fractional additional > investment relative to the original outlay, can be leveraged to > deliver (x)^4-n capacity -- with n yet to be determined? Must every > increment of the now vastly larger resource be priced as it would > have been assuming the "original" max cap? How much must the > "replacement cost" replace? The original (x) capacity? The as-yet > indeterminate (x)^n capacity? The originally anticipated/full > scarcity-based/monopoly-backed profits? > >> So everyone can extract profits for years, making out like bandits >> as they grow in to the >> excess capacity that was installed between 1999 and 2001, and they >> won't have a day of reckoning until they run out of capacity and >> find that they haven't been earning enough from their networks to >> service the debt they're going to need to take out to perform the >> next round of infrastructure upgrades. >> >> Example: You cannot seriously expect me to believe that the price >> of transatlantic connectivity actually reflects the cost of laying >> cables across the Atlantic. It defies common sense that a Gig-E >> tail from NYC to London is priced within an order of magnitude >> of a Gig-E tail from NYC to Boston. > > Once you get acquainted with the power of that ^n, you'll believe ;-) > Unfortunately, your location gives you few opportunities to > familiarize yourself. > >> Metered charging systems are, to me, evidence of a realization that >> the business model underlying much of the Internet's last five years >> is unsustainable. You guys might think they're a novel and >> unwelcome arrival at the moment, but give it a few years and we'll >> see what happens :-) > > If fine-grained metered pricing comes to the rest of the world, it'll > be because people roll over for it (you guys weren't given a choice). > If/when that happens, I'll be lobbying my local gov to turn over the > water infrastructure to me so I can replace it with > household Evian vending machines; and I'd recommend you all get in on > the ground floor in the air market ASAP. > Better be quick though, because the revolution will be just around > the corner... > > TV

1 0

Re: An Attempt at Economically Rational Pricing: Time Warner Trial
by Mark_Andrews＠isc.org 22 Jan '08

22 Jan '08

Another view from OZ. I've got a plan that is labeled as Unlimited (12G) Cable. 12G/M (12:00:00-23:59:59) 24G/M (00:00:00-12:59:59) (Disappears if the base 12G is used up) Once the 12G is used up it drops to 64k. Every 4th month is free as I also have a telephone with this provider. The cost actually dropped this month which is a free month. Previously the alotted transfer has gone up twice (from memory) with no fee increase. When I got the initial plan there wasn't a differentiation between AM and PM and the the allowance was for the whole 24 hour period. There are plans above and below this one. I can move up at any time in the month and have it take effect straight away (untested) and move down at the end of the current billing cycle. If I have the choice on when to do a high volume operation I do it in the AM (24G period). My father has a 200MB plan which gives him 600MB. His use is basically email only and software updates. The later is what can cause him problems if he doesn't manage his usage correctly. I've a friend that works for the provider in question and he is on their highest plan. He has 4 children and a wife who are all active users and does max out his plan on occasions. The big advanatge of these plans is that the cost is fixed even if I've used up all my alotted transfer. Mark

3 2

Re: Where is the nanog acronym glossary?, was Re: Lessons from the AU model
by ' =JeffH ' 21 Jan '08

21 Jan '08

FWIW.. http://nanog.cluepon.net/index.php/NANOG_Glossary =JeffH

1 0

Level3/GTEI well-known DNS down?
by randal k 21 Jan '08

21 Jan '08

Anybody know what is up/down with the GTE/Level3 name-servers used the world over on 4.2.2.1, 4.2.2.2? Suddenly started getting no responses from them for a couple of our side apps. I hope these servers stick around forever, they're kind of an Internet legend. Cheers, Randal

8 7

Hollywood's 'Untraceable': Fact or fiction?
by Stephane Bortzmeyer 21 Jan '08

21 Jan '08

Very interesting interview of an Hollywood consultant (and former FBI agent) about the facts in the movie 'Untraceable'. Among the many technical details, I note: > Q: Any other elements in the movie the naysayers may call you and > the writers out on as being technically inaccurate? > A: The IP addresses in the movie are not real, for obvious > reasons. You can't use real IP addresses, because it will point to > real IP addresses. It's similar to the 555 area code for phone > numbers in movies. Does the movie uses RFC 3330's 192.0.2.0/24? :-) http://www.networkworld.com/news/2008/011808-hilbert-q-a.html

1 0

RE: Cost per prefix [was: request for help w/ ATT and terminology]
by Neil J. McRae 21 Jan '08

21 Jan '08

Joe How many E1 customers can I plug into that device? You can take a cross section on this issue at various points. Think of the huge (sub RSP8) 7500 estates still in production, GSR Engine <5, 7609 <RSP720 and Flexwan1 and I think the early Juniper boxes - that can cope with todays size of routing table. The cost per prefix on those devices is very expensive indeed as the step upgrade driver for those devices is now the size o the table. We are all paying for it margin reduction... Regards, Neil -----Original Message----- From: Joe Abley <jabley(a)ca.afilias.info> Sent: 21 January 2008 00:15 To: William Herrin <herrin-nanog(a)dirtside.com> Cc: Patrick W. Gilmore <patrick(a)ianai.net>; nanog(a)merit.edu Subject: Re: Cost per prefix [was: request for help w/ ATT and terminology] A new cisco 2851 can be found for under $10k and can take a gig of RAM. If your goal is to have fine-grained routing data, and not to carry gigs of traffic, that particular router is perfectly adequate. If you're prepared to consider second-hand equipment (which seems fair, since it's not as though the real Internet has no eBay VXRs in it) you could get better performance, or lower cost, depending on which way you wanted to turn the dial. Sometimes it's important to appreciate that the network edge is bigger than the network core. Just because this kind of equipment wouldn't come close to cutting it in a carrier network doesn't mean that they aren't perfectly appropriate for a large proportion of deployed routers which take a full table. Joe

1 0

RE: Cost per prefix [was: request for help w/ ATT and terminology]
by Neil J. McRae 21 Jan '08

21 Jan '08

I'm sure they will make it work but there a long list of issues using PCs as routers that still applies capex goes to opex. -----Original Message----- From: Adrian Chadd <adrian(a)creative.net.au> Sent: 21 January 2008 02:23 To: Jeff McAdams <jeffm(a)iglou.com> Cc: nanog(a)merit.edu Subject: Re: Cost per prefix [was: request for help w/ ATT and terminology] On Sun, Jan 20, 2008, Jeff McAdams wrote: > A Linux box (*BSD, pick your poison) running Quagga or similar will do > the job at an extremely low price point. > > Yeah, again, not gonna want to pass gigs of traffic through it, but the > same concept does still apply. I dunno, the *NIXes seem suddenly interested in fixing up their internal network code to try and handle 10GE stuff "better". You might find that someone decides to drop in a crazy-looking intel e1000 NIC driver at some point thats tailored towards forwarding 64 byte frames at gige rate. The commercial guys hacking on FreeBSD routers/bridges keep telling me they've done it w/ custom coding on intel NICs on decent-looking PCIe hardware. It doesn't seem out of the realm of possibility nowdays. Adrian

1 0

RE: BGP Filtering
by Ben Butler 21 Jan '08

21 Jan '08

Hi Jason, Fantastic news, it is possible. We are using Cisco - would you be so kind as to give me a clue into which bit of Cisco's website you would like me to read as I have already read the bits I suspected might tell me how to do this but have guessed wrong / the documentation hasn't helped - so a handy pointer would be appreciated. Kind Regards Ben -----Original Message----- From: Jason Dearborn [mailto:jasondearborn@gmail.com] Sent: 15 January 2008 16:35 To: Ben Butler Subject: Re: BGP Filtering That's typically a function of your router software. Juniper, Force10, and Cisco all have support for this. Check your manual. On Jan 15, 2008 8:11 AM, Ben Butler <ben.butler(a)c2internet.net> wrote: > > Hi, > > Considering: > > http://thyme.apnic.net > > Total number of prefixes smaller than registry allocations: 113220 > !!!!! > > /20:17046 /21:16106 /22:20178 /23:21229 /24:126450 > > That is saying to me that a significant number of these smaller > prefixes are due to de-aggregation of PA and not PI announcements. > > My question is - how can I construct a filter / route map that will > filter out any more specific prefixes where a less specific one exists > in the BGP table. > > If my above conclusion is correct a significant portion ~47% of the > number of the prefixes in the table could be argued to be very > unnecessary at one level or another. > > Is such a filter possible easily or would it have to be explicitly > declared, any chance of a process the automatically tracks and > publishes a list of offending specifics similar to Team Cymru's Bogon BGP feed. > > As a transit consumer - why would I want to carry all this cr*p in my > routing table, I would still be getting a BGP route to the larger > prefix anyway - let my transit feeds sort out which route they use & > traffic engineering. > > Thoughts anyone? > > > Kind Regards > > Ben >

46 66

RE: An Attempt at Economically Rational Pricing: Time Warner Trial
by michael.dillon＠bt.com 20 Jan '08

20 Jan '08

> It's very true that the underlying cost of the technology is > falling. However, the cost of chip development is growing > rapidly as device size falls (tapeout costs are >$1M USD > today) and the market is somewhat limited, so the ability to > amortize your design cost across volume is somewhat limited. > To further limit the market by effectively constraining it to > your own internal consumption is going to further increase > this effect. Normally, when a device designer contracts out manufacturing of their design to the Asian factories, they constrain the factories to only produce devices for the designer themselves. If the designer of such comodity devices were to take an open-source approach, and allow the Asian factories to produce devices for anyone who wants them, then this constraint is removed. Of course, the price lowering still depends on lots of others adopting the same hardware so this only partially solves that problem. --Michael Dillon

1 0