May 1997 - Test - lists.nanog.org

Re: UUNET Press Release on Peering
by smd＠clock.org 13 May '97

13 May '97

"William Allen Simpson" <wsimpson(a)greendragon.com> writes: > In view of the difference, I would suggest that the web farms have a > case that UUnet should pay _them_ for the priveledge of accessing their > content! Anytime UUnet asks for a fee for peering, just tell them that > you really consider them unequal, and that they should > pay YOU! This would be a really good idea because it would give a strong push into investing in web caching technologies at the ISP level. Once that's been done, and access-level devices (customer-aggregating NATs and the like) can be gotten reliably to intercept queries directed towards anything other than a web caching hierarchy, in an intelligent engineering effort to keep traffic as local as possible when it's possible, the whole spurious argument that this is about web farms will go away. Moreover, a more general piecemeal settlements scheme would be much easier to put in place in an environment where the web farmers or small providers argue back that they ought to be paid by larger providers with huge numbers of end users connected through them. Finally, if you were UUNET and looking at things from a UUNET perspective, what's the difference between an entity exchanging traffic with them in one or two places across an FDDI switch and an entity exchanging traffic with them in one or two places across a Frame Relay switch? And wouldn't you start wondering when small providers will start paying their own customers for the privilege of talking with them? Sean.

1 0

RE: UUNET Press Release on Peering
by Peter Galbavy 13 May '97

13 May '97

On Monday, May 12, 1997 6:53 PM, Deepak Jain [SMTP:deepak@jain.com] wrote: > Strikes me that the terms "diversely routed" and "geographically diverse" > can be interpretted wildly. > > For most networks, I don't see a real problem dropping a couple extra > pairs of DS3s (assuming they were only at two NAPs to begin with). There is if you do not have any "domestic" US customers. See below. > If UUNet requires a DS3 interconnect at LINX, MaeWest, MaeEast and say > (for giggles) Australia -- AND a diversely routed DS3 backbone in the > U.S. that takes on an entirely different meaning than the above. ... Cool. I can do the LINX. :) Only problem here for me is that (miles * BPS) is a formula that does not work. My company has a very large customer base in the UK and is a direct competitor of UUNET PIPEX. Big deal you say ? Networks who are competitors always interconnect in the US you say ? This doesn't work in the big picture for us, sitting and the end of a very expensive international T3 line and only competing in one market area. (BTW We have not actually ever requested peering from UUNET outside of the AS1849 seen at the LINX - so this is all hypothetical and I am using UUNETs policy as an example of why some of these do not make sense in a global scene - yes I also know this is NANOG) Watch my hands on this one, and I will try to go slowly. Scenario (a) - Demon Internet install a trans-US DS3 backbone, all centred around New York, since this is where all our traffic flows throw. We have no US customers, so the network is wasted, and the DS3's get used as unidirectional pipes to the various NAPs. Anyone wanna share this network ? Scenario (b) - Demon Internet "pay" a direct competitor in their own market for transit. I hope I do not even have to spell out the wonderful sales calls our customers and prospects would get ? (BTW We *pay* MCI for >transit< services and have come to an arrangement with PSI - for those who may try traceroutes - but until the Concert merger is final, MCI is *not* seen as a competitor - and PSI we can deal with, since we are all nice people). Scenario (c) - Demon Internet approach a number of large NSPs in the US asking about customer-network peering only (not transit) at some mutually agreed single point - sharing the actual costs of delivery. In the US of course. Scenario (d) - Demon Internet approach same people as in (c) but for two points, one in our territory - the UK. This means the US NSP has an international backbone to worry about. Guess which one I like ? But I will settle for (c). (Oh (d) for those who didn't guess). Anyone out there who want to *peer* with us in Scenarios (c) or (d) who does not already do so, please let me know in private e-mail. Anyhow, the point I am trying to make is that it is very much pointless for companies like us to build a backbone when I do not have customers to use it, and also that paying a company in the US for services, when we have an "aggressive" competitive relationship in our own market is a non-starter. Regards, -- Peter Galbavy Demon Internet Ltd http://www.demon.net/

1 0

Re: NSI and competition to RIPE and APNIC
by Carl Oppedahl 13 May '97

13 May '97

At 10:22 AM 05/12/97 CDT, Stan Barber wrote: >Carl writes: >> And the outcome of all this affects every Nanog member. It affects every >> network operator, every ISP, every sysop. It affects the 1.2 million >> owners of the 1.2 million domain names in COM and NET and ORG. Every >> network operator, every SIP, every sysop, every domain name owner, should >> pay attention to all of this and speak up. > >Carl, I don't think there are 1.2 million domain name registrants as many >domains (like CyberPromo, for example) has registered multiple domains names. Nowadays, taking annual fees into account, the number of parties owning thousands of domain names is dropping quickly. Okay, so let's say it is a ten-to-one ratio (which it's not, I believe it is more like 1.5 to one) then the number of domain name owners would still be 120,000. And those 120,000 owners should still be speaking up, in the absence of which things may happen that they won't like later. >That being said, there is already an alternative in the county code domains. >There, the policies are different (though not necessarily better) than NSI. That is no comfort at all, none whatsoever, to someone who has invested years of time and sweat and energy and money in a COM domain name. If the answer is, "Hey, you don't like NSI's domain name policy? So give up your domain name and start over again in a two-letter domain!" then it is no answer at all. >That's not to say that the exising gTLD infrastructure does not need to >evolve. It must. I agree with you that those who are domain name registrants >in the current gTLDs have a stake in this evolution and should express a >thoughtful and constructive opinion on how this evolution should take place. >Unfortunately, some who have expressed opinions have chosen not to be >either thoughtful or constructive. I think some operators have been reluctant >to participate because of this. As I am sure you are also aware, the present structure of the .US domain is unworkable for any business that is located in more than one state, or for any business that thinks it might ever move from one state to another. There ought to be a .com.us, for example, and at present there isn't one.

3 2

Re: UUNET Press Release on Peering
by Sean Donelan 12 May '97

12 May '97

>Do you feel that UUNET's own network doesn't meet the criteria set out >in the press release? How so? > >Which of UUNET's peers who are able to continue to peer with them, would >you say do not follow the criteria set out in the press release? Actually I was thinking of a few other providers, but it is a common occurance among all providers with requirements for peering. MCI's requirement for a fully-meshed backbone. AGIS's requirement to connect to five exchange points, and already have peering agreements with many other providers. I haven't seen UUNET's actual peering requirements, just the press release, so maybe this is just a case of poor PR writing. But the press release does give some details about the backbone. It says DS3's or higher are used on "major" routes. Which begs the question, what does the backbone use on "minor" routes? Few providers have homogenous backbones. They usually have a wide range of circuit types, speeds, media. The definition of backbone, major, route, diverse, dedicated, etc are ratholes you go down when you venture past the 'public' interface between providers. I'm not trying to pick on a particular provider, just pointing out things you need to consider if you plan to write a requirements document. I don't like putting ratholes into requirements if they aren't really requirements. Last summer, you could substitute AGIS for UUNET, and had the same debate. In the end its not a techinical decision, its a business decision. I would really like to avoid the fiasco that happened with Western Union and the whole notion of "International Record Carriers" and "Record Carriers" that happened before the FAX machine made the teletype obsolete. -- Sean Donelan, Data Research Associates, Inc, St. Louis, MO Affiliation given for identification not representation

1 0

Re: UUNET Press Release on Peering
by Vadim Antonov 12 May '97

12 May '97

On Mon, 12 May 1997, Christopher Morrell wrote: >> Which of UUNET's peers who are able to continue to peer with them, would >> you say do not follow the criteria set out in the press release? >I know of at least one. I'm sure there are more. It seems, >that equity in traffic exchange may be an overriding >principle, although running many web-farms and small >backbone can certainly give you that. This is a serious misunderstanding of the economics of Internet. The "equity in traffic" is not important at all. What is important is equal share in carrying packet flows, which can be approximately measured in miles*bps. *That* costs money. Therefore, a small ISP dumping most of its traffic to others at one exchange point will cause long-haul backbones to carry its traffic most of the distance (no matter which direction the traffic goes), thus effectively subsidizing that small ISP. --vadim

5 5

Re: REMINDER
by Dimitri Krinos 12 May '97

12 May '97

In Pam's absence today, let me just mention that there are still a few rooms left at the conference rate for NANOG 10. The cut-off for the conference rate ($129) is 6:00 PM Est on May 15th. After that, the Hyatt's standard room rate for single occupancy is $205/night. If you do make reservations after the cut-off date, please still mention NANOG so that you get your conference documentation, etc. Dimitri Krinos IBM Global Services At 08:57 AM 5/9/97 -0400, Brad Miller wrote: >There are no rooms left at the discount rate. The current rate is $195.00. > Wow. > > >At 08:36 AM 5/9/97 -0400, Pam Ciesla wrote: >> >>The Hyatt Regency room rate cut-off date is May 15. The phone number >>is 1-800-233-1234 or (813) 874-1234, be sure to mention NANOG. The >>attendee list has been updated. If you are not on the list and should >>be please contact me. If you are planning on calling with payment info >>please note that I will be out of the office Monday, 5/12 & Tuesday, 5/13. >>Registration fee of $100.00 is due by May 28. Cost after the 28th is $120.00. >> >>xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx >>Pam Ciesla 313-936-0172 - phone >>Merit Network, Inc. 313-647-3185 - fax >>University of Michigan >>4251 Plymouth Road >>Bldg. I, Suite C >>Ann Arbor, MI 48105-2785 >>xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx >> > > > > >____________________________________________________________________________ >___________ >Brad Miller 703-448-4470 ext. 424 >Director of Operations bmill(a)cais.com >CAIS Internet > >

2 1

UUNET Press Release on Peering
by Kent W. England 12 May '97

12 May '97

UUNET issued a press release today on their peering strategy. It is located at: http://biz.yahoo.com/bw/97/05/12/wcom_x000_1.html The gist of the press release is that UUNET will peer with anyone who has "a national network with a dedicated, diversely routed DS-3 (or faster) backbone, and which will connect to UUNET at DS-3 or greater speeds in at least four geographically diverse locations." UUNET is being taken advantage of by "small regional ISP's and companies which provide ``web server farm'' services rather than Internet networking." and must protect itself from abuse of its backbone. Sounds like they are backing off, since this line of reasoning (modulo bandwidth and number of interconnects) has been standard procedure for all the national backbones since the end of the NSFNET. However, if you read the fine print about "a dedicated diversely routed DS-3 or faster backbone" that lets out all the "no money down instant backbones" that many smaller national ISPs and web farmers have built of late. Are they still insisting on NDAs before talking? --Kent ------------- Monday May 12 8:32 AM EDT Company Press Release UUNET details peering strategy Changing Internet economics prompt new policy FAIRFAX, Va.--(BUSINESS WIRE)--May 12, 1997--Stating that the economics of the Internet have changed radically in the past few years, UUNET Technologies Inc., the world's largest Internet service provider and a subsidiary of WorldCom Inc. (NASDAQ:WCOM), today detailed its policy regarding peering with other ISP's. The company said it will continue to peer with ISP's that can route traffic on a bilateral and equitable basis. However, UUNET will no longer accept peering requests from ISP's whose infrastructures do not allow for the exchange of similar traffic levels. ``A few years ago all ISPs were generally the same size and used each other's infrastructures to a more or less equal extent,'' said John Sidgmore, president and chief executive officer of UUNET. ``Today that situation no longer exists and consequently there are many cases where peering is not appropriate.'' One of the major principles of UUNET's policy is to peer with ISP's that operate a national network with a dedicated, diversely routed DS-3 (or faster) backbone, and which will connect to UUNET at DS-3 or greater speeds in at least four geographically diverse locations. Peering is an arrangement whereby ``peers'', or ISP's of similar size, route each others' traffic to destinations on their respective networks. Because the flow of data and use of infrastructure are anticipated to be approximately equal in both directions, no money changes hands in peering relationships. UUNET said it has received numerous peering requests from small regional ISP's and companies which provide ``web server farm'' services rather than Internet networking. These are, in effect, requests for UUNET to provide national and international data transport, as well as connectivity and support services, to companies which do not have the ability to provide similar services in return. Essentially, companies requesting peering in these situations are seeking to use UUNET's network for free, after UUNET has spent hundreds of millions of dollars to create its infrastructure. ``This is a purely economic decision,'' Sidgmore continued. ``We are 100 percent in favor of interconnection and won't deny access to anyone. However where the use of our respective infrastructures would clearly be imbalanced, we cannot reasonably be expected to provide our transport, route management and support resources at no charge.'' For those ISP's, or web server farms, seeking transport and route management services from UUNET, but which do not qualify as peers, UUNET offers wholesale connectivity services beginning at monthly rates of $2,000 for T1 connections and $6,000 for fractional T3 connections. UUNET currently provides such wholesale connections to several hundred ISP's. UUNET's network consists of a global backbone with multiple DS-3 (45 million bits per second) links on all major routes. It has an aggregate capacity well in excess of 5 gigabits per second. UUNET recently announced it was making a $300 million investment in its infrastructure which would quadruple dial capacity and raise the speed of backbone routes to OC-12 (622 million bits per second), dramatically increasing capacity. ...

6 5

Re: NSI and competition to RIPE and APNIC
by sob＠academ.com 12 May '97

12 May '97

Carl writes: > And the outcome of all this affects every Nanog member. It affects every > network operator, every ISP, every sysop. It affects the 1.2 million > owners of the 1.2 million domain names in COM and NET and ORG. Every > network operator, every SIP, every sysop, every domain name owner, should > pay attention to all of this and speak up. Carl, I don't think there are 1.2 million domain name registrants as many domains (like CyberPromo, for example) has registered multiple domains names. That being said, there is already an alternative in the county code domains. There, the policies are different (though not necessarily better) than NSI. That's not to say that the exising gTLD infrastructure does not need to evolve. It must. I agree with you that those who are domain name registrants in the current gTLDs have a stake in this evolution and should express a thoughtful and constructive opinion on how this evolution should take place. Unfortunately, some who have expressed opinions have chosen not to be either thoughtful or constructive. I think some operators have been reluctant to participate because of this. -- Stan | Academ Consulting Services |internet: sob(a)academ.com Olan | For more info on academ, see this |uucp: {mcsun|amdahl}!academ!sob Barber | URL- http://www.academ.com/academ |Opinions expressed are only mine.

2 1

NSI and competition to RIPE and APNIC
by Hank Nussbacher 12 May '97

12 May '97

I am not sure if this is the proper forum so if not, please send me to the right one. It appears NSI is establishing franchises throughout the world and the franchisees get commissions on domains and IP nets handed out. What is most interesting is IP net allocations. RIPE handles Europe, APNIC handles Asia Pacific and Internic used to handle everything else. It would appear that NSI is gearing up to challenge that premise by setting up frachisees all over the world to hand out IP blocks. How will that affect routing tables? Has IANA stated anywhere that Internic should *not* handle areas covered by RIPE and APNIC? Hank Nussbacher

6 10

large scale xDSL deployment?
by randy＠psg.com 12 May '97

12 May '97

Anybody aware of large scale xDSL ISP deployment, like hundreds of units? randy

6 5