One shouldn't necessarily believe any third-party web pages, documents, articles, or verbal statements about which networks any given network is peered with. Though I have seen many articles, web pages, and other tables that detail which networks are 'peering,' I have never seen an accurate representation of this type of data from a third party. It is difficult for a shared-medium (or direct) exchange point operator to keep this type of information up-to-date and accurate. ...and, more importantly, many of their customers consider this type of information confidential. To further complicate matters, many networks out there are mis-representing themselves as [cost-free-] peering with other networks when they are actually customers, or in some type of 'settlement' arrangement. As for your second question about the locations that networks interconnect, many networks consider this information sensitive as well-- though during your conversations with them, they will typically exchange that data with you-- though probably under a mutual NDA. Some choose to peer in locations where it is the most cost-effective between the two networks, while others peer in a smaller set of locations that make sense from a network hierarchy/topology perspective. The decision about whether to peer with a given network privately/directly versus at a public exchange point is often based on the amount of traffic to be exchanged, the network coverage area, and other such criteria. With a bit of knowledge, tools, routing table vantage points, and time, you should be able to find out most of what you want to know on your own. You might start by asking the operators of the exchange points (where you intend to connect) who their current customers are-- then you can contact those networks individually. Some exchange points will give you an "I want to peer with you" form and a list of e-mail addresses, so that you can mass-request. Don't get cook'ed by the wealth of misinformation out there. - jsb -- Jeff Barrows Director, Internetwork Engineering UUNET, an MCI-Worldcom Company
Date: Wed, 19 Apr 2000 09:20:03 -0500 From: Jeff Lentz <jeff@sbtek.net> To: nanog@nanog.org Subject: Peering Table Question
I apologize if this question is not relevant to your discussions, but Susan Harris, e-mail address srh@merit.edu, suggested that I ask it here...
The peering table that appears on the London Internet Exchange web site www.linx.net/peering.html...... is this something that can be generated for different peering locations? Also with this information would it be possible to create a table that lists companies and the cities that they are peering in? Any insight would be of great help to me thank you.
Once again any advice on how or where to obtain this info would be of great benefit to me. Thanks
Jeff Lentz
Jeff,
One shouldn't necessarily believe any third-party web
FWIW the LINX peering table is generated from the RA database. Nothing magic about it. So it's exactly as believable as the relevant two party's entries in the RA (i.e. not always believable). Where the two parties entries are out of sync, it adds a '?'. I believe LINX has put the code for it in the public domain. Please note it does *not* indicate the peering is at the exchange point concerned, nor that it is a peering, rather than a transit relationship. -- Alex Bligh VP Core Network, Concentric Network Corporation (formerly GX Networks, Xara Networks)
Though I have seen many articles, web pages, and other tables that detail which networks are 'peering,' I have never seen an accurate representation of this type of data from a third party.
amen!
To further complicate matters, many networks out there are mis-representing themselves as [cost-free-] peering with other networks when they are actually customers, or in some type of 'settlement' arrangement.
i contend that one can count the true tier one networks on the fingers of two hands.
As for your second question about the locations that networks interconnect, many networks consider this information sensitive as well-- though during your conversations with them, they will typically exchange that data with you-- though probably under a mutual NDA. Some choose to peer in locations where it is the most cost-effective between the two networks, while others peer in a smaller set of locations that make sense from a network hierarchy/topology perspective. The decision about whether to peer with a given network privately/directly versus at a public exchange point is often based on the amount of traffic to be exchanged, the network coverage area, and other such criteria.
With a bit of knowledge, tools, routing table vantage points, and time, you should be able to find out most of what you want to know on your own.
You might start by asking the operators of the exchange points (where you intend to connect) who their current customers are-- then you can contact those networks individually. Some exchange points will give you an "I want to peer with you" form and a list of e-mail addresses, so that you can mass-request.
Don't get cook'ed by the wealth of misinformation out there.
people should spend the tree and print jsb's message and frame it. randy
On Wed, 19 Apr 2000, Randy Bush wrote:
To further complicate matters, many networks out there are mis-representing themselves as [cost-free-] peering with other networks when they are actually customers, or in some type of 'settlement' arrangement.
i contend that one can count the true tier one networks on the fingers of two hands.
is a network tier one if they have settlement based peering? or is it just a network that sees no transit routes from any one company? christian
Wasn't that one of your top-10 lies of the internet? "We're a tier one isp..." joelja On Wed, 19 Apr 2000, Randy Bush wrote:
is a network tier one if they have settlement based peering?
no. because the folk they pay are not really PEERS, are they.
randy
-------------------------------------------------------------------------- Joel Jaeggli joelja@darkwing.uoregon.edu Academic User Services consult@gladstone.uoregon.edu PGP Key Fingerprint: 1DE9 8FCA 51FB 4195 B42A 9C32 A30D 121E -------------------------------------------------------------------------- It is clear that the arm of criticism cannot replace the criticism of arms. Karl Marx -- Introduction to the critique of Hegel's Philosophy of the right, 1843.
is a network tier one if they have settlement based peering? no. because the folk they pay are not really PEERS, are they.
As I understand it, peering means an agreement to share routing information about each other's networks. Traditionally peering has been free, but this changing. Simon -- Simon Lockhart | Tel: +44 (0)1737 839676 Internet Engineering Manager | Fax: +44 (0)1737 839516 BBC Internet Services | Email: Simon.Lockhart@bbc.co.uk Kingswood Warren,Tadworth,Surrey,UK | URL: http://support.bbc.co.uk/
At 01:02 PM 4/19/00 -0700, Randy Bush wrote:
is a network tier one if they have settlement based peering?
no. because the folk they pay are not really PEERS, are they.
Sure they are, or at least can be. AS1 used to claim that they wanted "settlement based peering" from people, and they said they would be 100% fair about it. If you peered with AS1 and had more than a 2:1 delta (to AS1 : from AS1), you paid $XXX per (something). If AS1 sent you more than double what you sent them, THEY would pay YOU. Now, are you going to tell me that AS1 is a "customer" of some network simply because they send that network more traffic? Besides, who really cares who is a customer of whom. All that matters is "how fast do my packets get there (and back), how many get dropped along the way, and how wide is the pipe", or "latency, bandwidth, and packet loss". Anything else is marketing, and best left off an OPERATIONAL mailing list.
randy
TTFN, patrick -- I Am Not An Isp - www.ianai.net ISPF, The Forum for ISPs by ISPs, <http://www.ispf.com> "Think of it as evolution in action." - Niven & Pournelle (Enable? We dunt need no stinkin' enable!!)
Errr, isn't 'settlement based peering' synonomous with 'usage-based transit' ?
not exactly. you get less routes and your routes do not propagate as far. hence, sometimes it is cheaper per megabit than normal transit. sometimes paid pseudo-peering is nice because, as the payee is really a customer, one does not have to be as formal about consistent application of peering qualifications as one does for true peering, when one has to presume that some day one will be explaining equitable treatment to the doj, ec, ... randy
Errr, isn't 'settlement based peering' synonomous with 'usage-based transit' ?
transit implies that the "peer" will show your routes outside their network, hence allowing traffic from some other network to "transit" their network on the way to your network. Whereas 'settlement based peering' is just charging for access to your/their network only. Of course, the fact that one can make 'usage-based transit' act like 'settlement based peering' both technically and financially with the use of bgp communities could blur the distinction. -mark
On Wed, 19 Apr 2000, Mark Kent wrote:
Errr, isn't 'settlement based peering' synonomous with 'usage-based transit' ?
transit implies that the "peer" will show your routes outside their network, hence allowing traffic from some other network to "transit" their network on the way to your network. Whereas 'settlement based peering' is just charging for access to your/their network only.
Does it? I don't agree. We often sell transit to customers, to which we only send routes of peers, and a discount.
Of course, the fact that one can make 'usage-based transit' act like 'settlement based peering' both technically and financially with the use of bgp communities could blur the distinction.
Actually, now that I think about it, I was wrong; usage based transit implies that the more you push/pull, the more you pay; settlement based peering could imply that, or that you pay for the delta between the two.
At 03:24 PM 04/19/2000 -0700, Mark Kent wrote:
Errr, isn't 'settlement based peering' synonomous with 'usage-based transit' ?
transit implies that the "peer" will show your routes outside their network, hence allowing traffic from some other network to "transit" their network on the way to your network. Whereas 'settlement based peering' is just charging for access to your/their network only.
I believe both of these statements to be false. 1. Many/most providers have the ability to prevent prefixes from being announced to other peers, setting a BGP community, etc. 2. Settlement based peering has more to do with traffic ratios than anything else. -Steve
On Wed, 19 Apr 2000, Steve Meuse wrote:
2. Settlement based peering has more to do with traffic ratios than anything else.
-Steve
You nailed it with that one. (According to our agreement with -NDA nixed name-) The question then becomes one of: Which end of the equation do you want to be on? The one sending traffic or the one receiving traffic? How does one determine who is at greater benefit from the peering relationship? John
On Thu, 20 Apr 2000, John Fraizer wrote:
The question then becomes one of: Which end of the equation do you want to be on? The one sending traffic or the one receiving traffic? How does one determine who is at greater benefit from the peering relationship?
If you are a web host you are generally willing to pay for bandwidth going towards the "net"....as your customers care if they can get to all of the "dialin" users. If you are a internet access provider you generally care more about bandwidth coming from the "net"....as your customers want to get to the web hosts. So assume that a primarily Web Hosting company peers with a internet access provider. The web host wants to pay to send traffic to the dialups and the iap wants to pay to recieve traffic from the web sites. The "willingness to pay" seems kinda one-directional here - the path from the web hosts to the iap. This whole settlement based idea is really screwy. If the model was that the people hosting the sites on the web paid for everything including access, then I could see settling based on bidirectional flows to a given "endpoint". Basically each section along the path to a given customer would get a small portion of the settlement. (you could argue the other way - that if the iaps were paying for the whole thing, the path the other direction would get it....). But that isn't the case. We live in a world where both the hosters and the accessors are paying for their access. Any settlement based in that system will be unfair as it is almost impossible to put a different value on each end of the link. Who's to say for a given AS whether inbound or outbound is more important? If someone (off list) could explain to me how flows in a given direction actually relate consistently to value, I'd appreciate it. Note that I am not talking about say a smallish ISP purchasing a "Customer Routes Only, no transit" link to a larger isp. That's a whole different ball of yarn. - Forrest W. Christian (forrestc@imach.com) KD7EHZ ---------------------------------------------------------------------- iMach, Ltd., P.O. Box 5749, Helena, MT 59604 http://www.imach.com Solutions for your high-tech problems. (406)-442-6648 ----------------------------------------------------------------------
are mis-representing themselves as [cost-free-] peering with other networks when they are actually customers, or in some type of 'settlement' arrangement.
i contend that one can count the true tier one networks on the fingers of two hands.
Howabout offering us your pearls and letting us in on who these mysterious select group of 'real' tier one networks are?
are mis-representing themselves as [cost-free-] peering with other networks when they are actually customers, or in some type of 'settlement' arrangement.
i contend that one can count the true tier one networks on the fingers of two hands.
Howabout offering us your pearls and letting us in on who these mysterious select group of 'real' tier one networks are?
the question is: does it make any difference today, other than from a marketing perspective? -b
I'm with Brett on this one... Is there any gain in advertising you don't have a back door if congestion / network failure occurs with a peer? Thinking back about all the work-arounds to get to specific peers that couldn't add capacity where they wanted because of long drawn out merger/acquisition activity makes me think a backdoor is wise and that customers would expect such professional planning. Sitting on customer calls I've heard that they don't want to go with network providers that might be transit free but are reputed congestion nightmares. Call it insurance against evolution... -Ren At 01:17 PM 4/19/00 -0700, brett watson wrote:
are mis-representing themselves as [cost-free-] peering with other networks when they are actually customers, or in some type of 'settlement' arrangement.
i contend that one can count the true tier one networks on the fingers of two hands.
Howabout offering us your pearls and letting us in on who these mysterious select group of 'real' tier one networks are?
the question is: does it make any difference today, other than from a marketing perspective?
-b
> Howabout offering us your pearls and letting us in on who these > mysterious select group of 'real' tier one networks are? By comparing competing claims, I think I've decifered the "tier" system: Tier 1 is the claimant's marketing department. Tier 2 is the claimant's customers and the claimant's competitors. Tier 3 is the claimant's customer customers, competitors' customers, and customers' competitors. -Bill
participants (16)
-
Alex Bligh -
Alex Rubenstein -
Bill Woodcock -
brett watson -
Christian Nielsen -
Forrest W. Christian -
I Am Not An Isp -
Jeff Barrows -
Joel Jaeggli -
John Fraizer -
Lauren F. Nowlin -
Mark Kent -
Randy Bush -
Rodney L Caston -
Simon Lockhart -
Steve Meuse