Talk:Peering

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Short definition of Tier 1[edit]

On Peering, changed definition of Tier 1 from "A network who sells to, or peers with, every other network on the Internet" to "A network who peers with every other network on the Internet". The new defintiion is derived from Tier 1 carrier, which seems a more accurate definition to me. -- Markmiddleton

No, that would be absolutely factually incorrect. Think for just a moment about what would be required to peer with every other network on the Internet. I operate a very large network, with just shy of 1,000 peers. There are probably a couple that have more peering than that. But certainly not all of the Tier-1s. Now imagine something _45 times larger_. Which is how you just tried to redefine it. -- Bill

Agree that this definition (and it's online right now) seems incorrect. What it's also got going against it, is that a Tier 1 provider can turn into a Tier 2 provider. And this happens without it doing anything at all. Once another country spawns a bunch of networks, and uses a Tier 2 provider to trunk/haul the traffic across oceans, then all the original Tier 1 providers aren't necessarily able to plug in all those new potential Tier 3 guys. Just a thought: How about amending it to "every other Tier 2 provider" ? Still not perfect, but better.Whophd 06:43, 14 September 2006 (UTC)

Huh? I think the definition of a T1 is they do not pay *anyone* in order to reach the entire Internet. Everyone else eithers pays them or exchanges it for free. So if a large US based provider has no presence in Asia but can reach all of asia via links which they are paid for or via free peering then they can still be a Tier 1 even if they don't talk to any provider in Asia.
Also the wording in that section is bad. "Network" is completely overused. Perhaps replace it with "organisation" or "autonomous system" to make things simplier. - SimonLyall 10:59, 14 September 2006 (UTC)

Hasn't this all become sort of academic and beside-the-point, since no "tier 1" network exist, by that definition? Is there any definition of "tier 1" which is not the empty set, anyway? Bill Woodcock 14:07, 27 January 2007 (UTC)

I have tried to avoid the problem above by calling it "overt settlements", or settlements where cash changes hands specifically to pay for things like ratios or inadequate traffic volumes. Although this is obviously not perfect, it at least gives us a definition which will not invalidate everything "known" to date. Patrick W. Gilmore 04:35, 29 June 2008 (UTC)

Clarification[edit]

ICAIS description relies on implicit conclusions and inferences drawn by an educated reader. Even then, the thing is presented from an almost theoretical, save the terminology, and politically based view, which gives no detail and makes it difficult to form an opinion. After reading it several times over, I concluded that: International Charging Agreements for Internet Service is viewed from an economic perspective. Adoption would decrease costs for those least able to pay and would arbitrarily place economic burden on those who bear no moral or legal responsibility to remedy fabricated injustice. AwesomeMachine 17:47, 31 May 2011 (UTC)


please define MED, and other acronyms. link to other wiki-pages if applicable. do not use an acronym without spelling it out the first time used.

Someone has since done this. -- Beland 23:56, 16 May 2006 (UTC)

Peering dos and don'ts[edit]

Some of the general guidelines to successfully engaging in peering include:

  • Do: Respond promptly to NOC emails/calls.
  • Do: Have administrative, technical and NOC contact information ready on any peering request.
  • Do: Maintain a database of traffic to existing peers to assist with peering upgrades as traffic increases. Regularly examine how much traffic you do to your peers at a set interval so you can prevent any capacity issues.
  • Do: When approaching a network to peer, have traffic statistics ready.
  • Do: Treat your peering sessions like any other production circuit. Just because it is free does not mean you can shut it down at any time.
  • Do: Maintain a peering "web site" which includes your locations, IP addresses, ASNs, contact information (administrative, technical, NOC) and policies.
  • Do: Use www.peeringdb.com to share your peering information with other networks.

If you have to reduce/increase your prefix count, please let your peers know when, how long, and if possible, why.

  • Do not: Leak prefixes to your peers.
  • Do not: Point non-advertised routes or default routes to peering nexthops.
  • Do not: Sell specific peer routes to customers (known as selling nexthop).
  • Do not: E-mail bomb existing peers, or peers who do not respond to you.
  • Do not: Advertise inconsistent prefixes at multiple peering locations.
  • Do not: Expect the other peer to honor your MEDs
  • Do not: Lie about your network size, network capacity, or traffic to a potential or existing peer.
  • Do not: Setup eBGP sessions via public peering to networks that have not agreed to peer with you.
  • Do not: Attempt to modify BGP origin codes to force cold-potato routing.

I removed the above because it is normative, which is not NPOV. -- Beland 23:55, 16 May 2006 (UTC)

I disagree. Those are an accurate description of accepted best practice in the industry. That is, a description of what is, not what should be. Though their description is not the best, shouldn't be do and don't lists but practices generally considered helpful and unhelpful. Perhaps that distinction is why they appeared to be recommendations rather than actual accepted practice. Jamesday 19:20, 5 September 2006 (UTC)

I keep removing the "Peering Ecosystem" document because it's a paper designed to be a marketing document. Martin.hannigan 11:02, 16 July 2006 (EDT)

Clarification[edit]

With the following line:

Peer directly with that network, or with a network who sells transit service to that network, or

If A pays B, B pays C and C has a peering with D, does it mean that D can send traffic to A through B and C? Thanks 89.180.64.93 05:54, 10 May 2007 (UTC)João