- Test - lists.nanog.org

RE: MFS WorldCom/WilTel/LDDS
by Peter Ford 03 Sep '96

03 Sep '96

All of them do, to my knowledge. It's Metcalfe et al who are calling for change in the direction of resource reservation and per-connection charging. --vadim Vadim, Metcalfe certainly is not the first, or the last, to call for "reservations" and last time I looked a lot of the work in the IETF is centered on this area. This does bring up the interesting point of will/should the Internet business model change to support reservations. Let's assume some kind of "promise" beyond the current best effort service is offered. Do you really believe this will not be based on usage? peter

1 0

Strange Routes
by Vadim Antonov 03 Sep '96

03 Sep '96

Enjoy: % traceroute www.unitedmedia.com traceroute to unitedmedia.com (204.240.178.10), 30 hops max, 40 byte packets 1 140.174.86.10 (140.174.86.10) 137.582 ms 125.193 ms 126.528 ms 2 gw-tlg (140.174.86.2) 130.345 ms 126.407 ms 158.074 ms 3 tlg-cust-link.tlg.net (140.174.199.2) 134.497 ms 126.107 ms 127.23 ms 4 ln3_gw1-sj-tlg_mv.tlg.net (140.174.101.5) 204.124 ms 255.635 ms 194.128 ms 5 mae-west.tlg.net (198.32.136.22) 183.47 ms 223.452 ms 206.021 ms 6 905.Hssi3-0.GW1.SCL1.ALTER.NET (137.39.133.89) 204.209 ms 266.514 ms 278.43 ms 7 Fddi6-0.SR1.SCL1.Alter.Net (137.39.19.36) 222.749 ms 330.877 ms 490.011 ms 8 Hssi2-0.BR1.NUQ1.Alter.Net (137.39.103.6) 380.718 ms 234.135 ms 436.153 ms 9 mae-west.psi.com (198.32.136.23) 473.437 ms 287.215 ms 305.98 ms 10 ne.sc.psi.net (38.1.3.1) 500.152 ms 367.262 ms 430.873 ms 11 rc8.northeast.us.psi.net (38.1.21.8) 370.318 ms 509.708 ms 386.508 ms 12 38.4.218.4 (38.4.218.4) 461.314 ms 443.618 ms rc8.northeast.us.psi.net (38.1.21.8) 507.873 ms 13 unitedmedia.com (204.240.178.10) 447.827 ms 335.4 ms * % --vadim

3 2

Re: Access to the Internic Blocked
by Geoff Huston 03 Sep '96

03 Sep '96

Vadim writes, >On itself, LSRR is a godsend to hackers (i can think of about >a dozen of very nasty attacks using general LSRR). The only >useful application for it is traceroute. > >Why don't router vendors provide an option to turn it >off for everything but ICMP ECHO? Personally I find this an excellent suggestion. g

3 2

Re: Agenda for next NANOG
by Guy T Almes 03 Sep '96

03 Sep '96

Randy, There is a cluster of closely-related variants on the issue you raise: [] to what extent does a given exchange point (NAP/MAE/etc) constrain the performance that a user sees (in what a user thinks of as end- to-end). For example, an FTP could flow at 800 kb/s for a given pair of users, except that MAE-north is congested, so the FTP can only flow at 400 kb/s. [] to what extent does a given exchange point constrain the performance that a provider sees (in what a provider thinks of as end-to-end). For example, a given pair of backbones could sustain 20 Mb/s over a private interconnect with acceptable packet loss, but can only sustain 10 Mb/s over the Altoona NAP. >From different perspectives, each of these notions of end-to-end has meaning. Would you want to consider one or the other or both in the panel? -- Guy At 01:11 PM 8/30/96 PDT, Randy Bush wrote: >>> Analysis of Actual End to End Performance accross the NAPs/MAEs >> An excellent topic, to be sure, but how do you propose that this be >> measured? > >And there's a subject for a NANOG panel in itself. > >I can think of some interesting experiments that would involve cooperation >of multiple peers. But there are folk far better based in measurment than I >who might suggest some fun stuff. Guy, Steve, ..., this is your cue. > >randy >

2 1

Re: Agenda for next NANOG
by Guy T Almes 03 Sep '96

03 Sep '96

Paul, The relevant BMWG work here is within the IPPM effort. The email list is ippm(a)advanced.org with the unsurprising ippm-request(a)advanced.org to join/unjoin. -- Guy At 01:29 PM 8/31/96 -0400, Paul Ferguson wrote: >At 01:11 PM 8/30/96 PDT, Randy Bush wrote: > >>>> Analysis of Actual End to End Performance accross the NAPs/MAEs >>> An excellent topic, to be sure, but how do you propose that this be >>> measured? >> >>And there's a subject for a NANOG panel in itself. >> >>I can think of some interesting experiments that would involve cooperation >>of multiple peers. But there are folk far better based in measurment than I >>who might suggest some fun stuff. Guy, Steve, ..., this is your cue. >> >>randy >> > >... And it might even be nice for someone who is active in the >IETF Benchmarking Methodology WG (bmwg) to participate. > >[ref: http://www.ietf.org/html.charters/bmwg-charter.html] > >- paul >

1 0

Re: Agenda for next NANOG
by salo＠msc.edu 03 Sep '96

03 Sep '96

> Date: Fri, 30 Aug 1996 15:59:23 -0700 (PDT) > From: Michael Dillon <michael(a)memra.com> > To: nanog(a)merit.edu > Subject: Re: Agenda for next NANOG > [...] > What would it take to set up a couple of Alphas, routers and zero-mile > T1's in a portable testing kit to go around from NAP to NAP and run tests? > [...] > Such > machines could run a full suite of tests including FTP'ing large files, > downloading complex web pages (i.e. multiple images) as well as the lower > level things like ping tests. > > This kind of test would provide useful numbers that customers can > understand as well as point out problem areas that a NAP operator might > need to investigate. And with co-operative NAP peers, the same test kit > could be used with T1's out to various locations that feed into the NAP so > as to run the same tests across the peer's routers and lines. > > Would this kind of testing reveal any useful information that could not be > gotten from examining router stats? Actually, it would be nice if this "portable test kit" were actually an optional board for a router. You could, for example, stick the "portable test kit" board in a router and then test at will. We also tried to explain to our ATM tester vendors that we wanted ttcp implemented in the ATM tester. Some understood, some didn't... -tjs

4 4

RE: MFS WorldCom/WilTel/LDDS
by Jim Browning 03 Sep '96

03 Sep '96

>From: Sean Donelan[SMTP:SEAN@SDG.DRA.COM] >Sent: Monday, September 02, 1996 3:22 PM >A current problem Internet is line capacity only comes in a few different >sizes; small (DS0), medium (T1), large(T3), and extra-large (OC3c). >One reason ISDN is extremely popular is its just a bit bigger than >a DS0, and a lot less than a T1. Because the next step is so large, >customers tend to stay with their current size long after it stops >fitting their needs. It's possible to provision ATM circuits at pretty much any capacity a customer wants, from 1Mbps to 155Mbps, without having to swap-out hardware in order to bump up the speed. Of course this also makes new services available which would otherwise be impossible to manage (single event bandwidth increases, anyone?) >So-called "usage-based" pricing is just a klunky work around to the too >big a step between sizes problem. Customers want something just a bit >bigger than a T1, but don't want something 28 times bigger. Usage-based >pricing lets customers use something just a bit bigger even though the >provider uses a massive pipe to deliver it. This is frequently true, although there are specific situations where usage based pricing is attractive for its own sake, for instance as backup for a fixed price circuit. >I suspect when Cisco's 11.2 traffic-shaping hits the streets, you >will be able to buy more "fixed-rate" Internet connections at the >in-between line-speeds providers previously only offered on >usage-based pricing plans. Sure, some providers may resist initially. >But someone will break ranks, and the others will eventually follow. Those ranks have already broken, at least in territories we serve, and there was no need to wait for new software... ;-) -- Jim Browning <jfbb(a)atmnet.net> P.S. Will it help avoid the IP <==> ATM religious battle I may have just started if I lie and say I am unsubscribing for the next week or so?

1 0

Re: MFS WorldCom/WilTel/LDDS
by Sean Donelan 03 Sep '96

03 Sep '96

>I think you are confusing dialup and leased line capacity planning >models. They are inherently different. If I have a POP in City A They aren't inherently different. You run into the same problems at every level. 8 28.8Kbps dialup modems into a terminal server hanging off a 28.8Kbps line (as some ISPs have been known to do) has the same problem as 8 T3 customers off the same T3 provider line (as other ISPs have been known to do). >The trend is for flat rate dialup and usage based rates for leased >line. Those ISPs that ignore that trend will not last. The kicker In industries with competition, customers have shown a preference for fixed rates over variable rates. Avis, Hertz, etc tried several times to introduce milage charges on car rentals in the US. Some car rental company always breaks ranks, and "unlimited" milage car rentals return. A current problem Internet is line capacity only comes in a few different sizes; small (DS0), medium (T1), large(T3), and extra-large (OC3c). One reason ISDN is extremely popular is its just a bit bigger than a DS0, and a lot less than a T1. Because the next step is so large, customers tend to stay with their current size long after it stops fitting their needs. So-called "usage-based" pricing is just a klunky work around to the too big a step between sizes problem. Customers want something just a bit bigger than a T1, but don't want something 28 times bigger. Usage-based pricing lets customers use something just a bit bigger even though the provider uses a massive pipe to deliver it. I suspect when Cisco's 11.2 traffic-shaping hits the streets, you will be able to buy more "fixed-rate" Internet connections at the in-between line-speeds providers previously only offered on usage-based pricing plans. Sure, some providers may resist initially. But someone will break ranks, and the others will eventually follow. Customers like fixed rate prices. And as bank mortgage rates show, customers are often willing to pay more for a fixed-rate price than a variable-rate price. As long as there is competition, ISPs that ignore that trend, may end up like the car rental places that charge by the mile. -- Sean Donelan, Data Research Associates, Inc, St. Louis, MO Affiliation given for identification not representation

1 0

Re: MFS WorldCom/WilTel/LDDS
by Vadim Antonov 02 Sep '96

02 Sep '96

Hank Nussbacher <hank(a)ibm.net.il> wrote: >>On Thu, 29 Aug 1996 01:47:38 -0700 Vadim Antonov wrote: >>Generally speaking, at least 70% of users already have metered >>access of that kind (in form of per-hour connect time charges), >>so in this sense Internet is metered already, and was like that >>for years. Obviously that's not "metering" Metcalfe et al >>are calling for. >I think you are confusing dialup and leased line capacity planning >models. They are inherently different. If I have a POP in City A >with a T1 to my backbone, I pretty much know how many dialups I can >support. If you do even minimal statistics on leased lines you'll know what customers use and how their utilization grows over time. When you have a provider which runs thousands of leased lines it becomes no more of guesswork than dialup. >Just look at the example of the NANOG poster who had to move ISPs >since his thruput was high enough to double his costs so he moved >to a flat rate ISP. Those flat rate ISPs will quickly load up with >these thruput heavy users and in the end either lose money every >year or go bankrupt. It is no different from local telco pricing in US where you can get metered access if you're low-volume, or flat-rate if you're willing to guarantee payment for some usage. As for "losing money" -- i've seen SprintLink's balance sheets. It's not going bankrupt any time soon. In fact, the lack of differentiation between classes of customers is what made Sprint one of the top players in the field -- most small ISPs liked the pricing. I guess your view is different because of difference between pricing of leased lines. If you have a hugely expensive international private line, you have to sell its capacity for prices which make customers unlikely to buy flat-rate. It's not like that in US. >The trend is for flat rate dialup and usage based rates for leased >line. Sorry, this does not correspond to what i've seen. People who buy a T-1 usually have enough traffic to fill it. Faster WWW access encourage faster browsing, that's it. >Those ISPs that ignore that trend will not last. The kicker >is Cable/VSAT. A consumer that is offered 2Mb-10mb to the home via >cable or DirecPc (bidirectional) for say $50 flat rate will >single-handedly be able to overrun the backbone line of the POP. This is back to the situation when "exterior" link is small as compared to "interior". In this situation usage-based pricing makes a lot of sense. Until first customers get slapped with $1k bills for gee-whiz automatic loading of some movies off the Web. The rule of backbone engineering is that you have to have some ratio (usually about 1:3) of backbone capacity to customer access capacity. If that ratio is 1:1000 (as in 10Mbps customer access to everybody in neighbourhood, with OC-3 backbone) it is bound to create problems, to say it mildly. That's why xDSL and cable Internet customer access remains largely a joke. >But when I specify usage based rates I am referring to a simple >meter like the gas company, not a complex meter like the phone >company. Then we're in agreement on that. However to introduce pricing sensitive to resource allocation in backbones, like Metcalfe advocates you will need telco-style end-to-end billing. It is not access capacity which is in shortage, it is backbone capacity. >As someone else stated, telco's are approaching 50% costs >for billing due to their overwhelming complexity between local, >long distance, international, time of day, length of call, services >used, etc. It's not complexity of computations, it is the fact of collection and archiving per-call data which makes that expensive. (BTW, if you charge per connection you'll have to keep records about that connection for at least half a year, as an evidence in case of payment disputes -- if you don't do that anybody will simply be able to say "i didn't make that call"). >Gas and most electric companies charge a flat rate per >usage, no matter what time of day you use it. Ah, again, they can accumulate resources to smooth out demand. (It's a bit harder with electricity, but there's such thing as accumulating power stations). >Their billing costs are far less than telcos. They don't have to generate itemized bills, and their entire usage records on you contain 48 bytes per year. >ISPs, in order to survive, had best not >follow the telco model of billing complexity, but instead follow the >utility company model of simple billing. KISS. All of them do, to my knowledge. It's Metcalfe et al who are calling for change in the direction of resource reservation and per-connection charging. --vadim

2 1

Re: MFS WorldCom/WilTel/LDDS
by Rbohn＠ucsd.edu 01 Sep '96

01 Sep '96

I agree with the thrust of this message (that flat rates on leased lines will run into what economists call "adverse selection" problems, while for dial-up there are inherent limits that make flat rates bearable). However a minor quible/question: there is an additional problem with badly written applications, such that they can SEND more than 28.8kb/sec even though the receiver can't get it (the difference ends up getting thrown away at the ISP). Thus a single 28.8 user can consume much more than that on incoming capacity, if they watch a lot of badly written real time video for example. Since the user will see very poor quality in this scenario (most packets are not being received), it does not seem likely to last long. But EUNet reported this problem, for example. Does anyone know how MBONE handles this issue? (Does it attempt to send all packets to every recipient, whether or not they can get them?) Real Audio? Roger Bohn At 6:41 AM 9/1/96, Hank Nussbacher wrote: ... > But I do not have to worry that a lone user running >CuSeeMe (or even SmellU/SmellMe (see realaroma.com/realaroma.cgi) >:-)) since the worst they can do is max out at 28.8kb on the T1 >line. > >On the other hand, if I have a number of leased line customers >connected to that T1 POP, some are 64kb, some are 256kb and some are >T1, then my model alters since the T1 users can indeed max at the >full speed of the line and others can utilize 15% of the line as >compared to 1% for a dialup user. Yes, the statistical model shows >they will even out, some will use 1%, some 10% and others will use >80% but in the end it all evens out. But that lone wolf T1 client >can indeed saturate my line, especially if it is a garage ISP with >1000 users from the neighborhood offering flat $9.95 and having >a modem bank of 96 modems. Or a small startup with every user >doing RealAudio during the day. > >Just look at the example of the NANOG poster who had to move ISPs >since his thruput was high enough to double his costs so he moved >to a flat rate ISP. Those flat rate ISPs will quickly load up with >these thruput heavy users and in the end either lose money every >year or go bankrupt. > >The trend is for flat rate dialup and usage based rates for leased >line. Those ISPs that ignore that trend will not last. rest of message omitted

1 0