January 1997 - Test - lists.nanog.org

Re: peering charges?
by Sean Donelan 26 Jan '97

26 Jan '97

>Or...since the business model of many (but not all) major web sites is >related to advertising, they should pay isp's for access to their audience >(client base of isp). It is the audience that makes the web site more >valuable....not the other way around. Another fine com-priv debate rehashed... So if UUNET's web hosting division wants access to AOL's client base, the payments should flow which way? Personally, I think it would be great, because something like 85% of our 'paid' traffic with UUNET is to/from UUNET hosted web sites. When an ISP has both consumers and producers, or buyers and sellers, the value flows both directions. You get inequities when there is an pricing inbalance, e.g. international telephone rates. Competitive markets tend to force inbalances to zero. The problem is in a free and open competitive market, any settlement scheme with net difference between providers, will result in a new provider entering the market whose business will maximize their net settlements. International call-back telephone services is one example. Zero net settlements is actually a great barrier to entry for new providers. -- Sean Donelan, Data Research Associates, Inc, St. Louis, MO Affiliation given for identification not representation

1 0

peering charges: ANSWERED
by Dave Curado 26 Jan '97

26 Jan '97

Just thought I'd post a follow up... It sounds like someone has spooked my CEO. While yes, we've all heard the idea of settlements before, it doesn't sound like it is any closer to reality then it was six months ago. To add my own $0.02 to the discussion regarding the question... It seems like a simple "supply vs. demand" issue to me. I would not be surprised if a large provider suddenly said, "you know, people *want* to peer with us... I bet we could charge for this as a service...", and some of us would pay, provided the fee could be cost justified. In any event, Thanks for the feedback on this question. davec

1 0

Pointer to AS Statistics; ISP/BGP training
by Howard C. Berkowitz 26 Jan '97

26 Jan '97

I'm feeling stupid about this, but I seem to remember that the information I am looking for is already in a MERIT or some other report, and I can't remember which it is. Disclaimer, on which I expand below: I do have commercial reasons for asking this, but I also hope to make the commercial results into something generally useful. Specifically, I am trying to determine some counts for: AS that generate advertisements only for their own AS. The assumption here is these will be either multi-homed enterprises or non-transit ISPs. This could be refined with a verification that if AS1 only advertises itself, it is actually multihoming only if the path AS1 ASx is advertised by at least 2 other AS. AS that provide transit. Ideally, I'd like to separate this into "large" and "not-large," the distinction being those that require scalable IBGP technologies beyond full mesh. Why am I doing this? Crass commercial reasons, but involving information that I am quite willing to share with NANOG, and which I believe is of benefit to the community. As many of you know, I work for an internetworking training organization. On my own, I have drafted some BGP and related courses, aimed at getting exterior routing beginners to the journeyman level. With the growth of ISPs, both expansion in existing ones and new entries to the market, some of which we may feel are clueless, I believe there is a need for education beyond that which is feasible as on-the-job. Before progressing further, I need to establish to my management that there is a training market here, and to the BGP-using community that it is possible to do meaningful training in the basics, beyond a brief lecture-only model but not to a level that builds a skilled exterior routing architect. I do hold personal copyright to the first versions of the courses. While obviously I'm looking for return on investment, as would my management, I personally believe it is of value to make such training available. I would, in fact, be hoping to find part-time instructors among the established exterior routing community. I'll be happy to share the curriculum model I have developed, and feedback is more than welcome about this whole idea. Fundamentally, it is a personal goal that may become commercial. Howard Berkowitz

1 0

Re: peering charges?
by Vadim Antonov 26 Jan '97

26 Jan '97

Jonathan Heiliger <loco(a)isi.net> wrote: >Large ISPs should probably be interested in access to content, without it >their users could find the Internet a very boring place. Not only that. They are also interested in peer's customer population gaining access to content (and ads) published by their own customers. --vadim

2 1

peering charges?
by Dave Curado 26 Jan '97

26 Jan '97

Curious thing, the CEO of my company called me the other day and was worried about who we were peering with. It seems someone put a bug in his ear that the larger providers were about to start charging some of the smaller providers for bilateral peering. I haven't heard anything about that at all... so either a) the ceo is mistaken b) I've missed some important annoucement recently either is possible. Has anyone else heard anything about this? Thanks, davec

4 3

Re: peering charges?
by Matt Harrop 26 Jan '97

26 Jan '97

At 02:38 AM 1/26/97 -0800, you wrote: >Jonathan Heiliger <loco(a)isi.net> wrote: > >>Large ISPs should probably be interested in access to content, without it >>their users could find the Internet a very boring place. > >Not only that. They are also interested in peer's customer population >gaining access to content (and ads) published by their own customers. > <My point of view may be very different from most of the list since I'm in Toronto, Canada> From my point of view (local ISP) I simply look at the cost versus performance. We're connected to our primary transit provider (UUNet Canada) via 100MB fibre, and we pay for average bandwidth usage. If paying for a non-transit (peering only) T1 into Sprint Canada ends up giving me equal or better performance for less money than paying for another 1.2Mbit of usage on my UUnet connection I'll be happy to pay. Whether I should be paying Sprint Canada for peering, they should be paying me, or we shouldn't charge each other a dime isn't relavent. When it comes right down to it either I'll peer or I won't based on the economics... If they want to charge an unreasonable sum (defined as more than it would cost me to send the traffic through UUnet onto Sprint Canada) then I won't peer with them. In which case they lose the revenue that I might have been willing to provide them. I'm still happy because I'm getting the most bandwidth for the least dollars. My case may well be fairly unusual in that we have big bandwidth (for a local ISP) but service only one area code, so we don't have to fund any long haul network of our own. Just our luck to be in the single largest (by population) local calling area on the planet ;) My $0.02.

1 0

Re: peering charges?
by Vadim Antonov 26 Jan '97

26 Jan '97

> You can tell this list is populated only by Real Engineers. Ten messages > and counting on Saturday night. :-) I'm doing my talk-to-capitalists routine more often than any code hacking nowadays... --vadim

1 0

Re: peering charges?
by Dave O'Shea 26 Jan '97

26 Jan '97

From: Vadim Antonov <avg(a)pluris.com> > >Since some of the larger vendors (Cisco mostly) has introduced accounting > >features into their software settlements could start any time. > > a) the accounting was there for years, so what .. But the huge glut of "I wanna be an ISP too!" guys operating a Cisco 2500 out of their garage was not. A large number of ISP's simply don't know what they're doing; witness the universal broadcasting of RFC1597 addresses. > > b) a 100-byte packet travelled from provider A to provider B. Should A pay > to B or vice versa? It's the golden rule - "He who has the gold, makes the rules". Not that the idea isn't without problems.. But seeing peering procedures formalized would make life easier, even if it cost a few bucks. Being a fairly small start up, I know that the odds of UUnet cutting me a check every month are between slim and none. You can tell this list is populated only by Real Engineers. Ten messages and counting on Saturday night. :-)

1 0

Re: operational problems at mae-west?
by Paul Ferguson 25 Jan '97

25 Jan '97

At 09:18 AM 1/25/97 -0700, Brett D. Watson wrote: > > i really don't understand why it's not in mainline though (like >11.2(2)). > They will be in the 11.2_F train. - paul

1 0

Re: operational problems at mae-west?
by Brett D. Watson 25 Jan '97

25 Jan '97

> At 02:48 PM 1/24/97 -0700, Brett D. Watson wrote: > > >2. ios version 11.1.8 early deployment (which i'm told by isp-team i > >shouldn't run unless i *really* feel the need) > > > > Don't know why there should be any issues here; 11.1(x)CA > contains the knobs, is running in several other large production > networks, and is readily available. > yes,i have to correct myself. i just found the knobs in 11.1.7ca1. i shouldnt' have believed them when they told me it wasn't there :) i really don't understand why it's not in mainline though (like 11.2(2)). -brett

1 0