Peering versus Transit

Jeff Young young at mci.net
Mon Sep 30 21:17:04 UTC 1996


because ds-3's to naps are finite resources and cost money.
why would you assume that any provider wants to burn his ds-3
by taking traffic at a nap when he has better connectivity 
to your transit provider?  in fact, he may not even peer with 
your transit provider at the nap.  

why am i not free to arrange traffic flows between my backbone
and others as i see fit?  mci and sprint have arranged six ds-3's 
between their respective backbones.  if your transit provider is
sprint, i don't want your traffic to me by way of a nap.  if you 
give it to me at a nap, you deserve what you get.

one would think, in my case, that a ds-3 to a nap would cost
me more than a direct ds-3 connection to the XXX backbone.
(assuming that XXX is the transit provider).

Jeff Young
young at mci.net

> Date: Mon, 30 Sep 1996 00:40:49 -0700 (PDT)
> From: Bill Woodcock <woody at zocalo.net>
> Message-Id: <199609300740.AAA01931 at zocalo.net>
> To: avg at quake.net, barney at databus.com, nanog at merit.edu
> Subject: Re: Peering versus Transit
> Sender: owner-nanog at merit.edu
> Content-Type: text
> Content-Length: 1837
> 
>         > There are two ways to have packets go...
>         > adding bogus static or whatever routes
>         > or by pointing default.  Both are malicious.
>     
>     WRT the latter, I completely agree that pointing a default at anyone
>     whom you're not buying transit from is theft, and absolutely beyond
>     condonement.
>     
>     WRT the former, I simply cannot fathom, and no one other than Sean has
>     yet presented an argument explaining why it's malicious to deliver a
>     packet to its addressee's ISP.  Why should I, as an ISP, not prefer
>     that all other ISPs deliver packets to my customers as quickly,
>     efficiently, directly, and inexpensively as possible?  Why should I
>     prefer a more expensive or less reliable route, or expect any other
>     ISP to do so?
>     
>     I realize that this is about the hundredth time somebody has asked
>     exactly that question, but people are just going to keep asking until
>     there's a convincing reason, or people stop suggesting that other
>     people use less-efficient paths.  It is, after all, an obvious
>     question.
>     
>         > Example, please, when somebody conforming to the stated policies
>         > was denied peering?  (Plase note that the process... may be
>         > rather lengthy...
>     
>     Okay, it's _widely rumored_ that it may be difficult to establish new
>     peering sessions with some large ISPs, at the moment.  :-)  But this
>     again distracts from the question at hand, since you assume that
>     "stated policies" should institutionalize unequal relationships.
>     Assuming that skirts the argument, just as nonsequiturs about default
>     routes do.
>       
>                          -Bill 
> 
> ________________________________________________________________________________
> bill woodcock  woody at zocalo.net  woody at applelink.apple.com  user at host.domain.com






More information about the NANOG mailing list