On 24 May 2004, at 20:18, Paul Brislen wrote:
TelstraClear tells me the reason for the introduction is that some customers were happily peering with TC in Wellington and shipping data to Auckland over TC's backbone without having any formal (ie paid) relationship with TC for the carriage of said data.
If that snippet is describing shipping traffic between an Auckland peer of TCL and a Wellington peer of TCL over TCL's network, then there are technical solutions to this problem which don't require TCL to de-peer (elements of which have been implemented in the past in one or more of AS 7714, AS 4763 and AS 4768). I haven't spent much time delving through the impressive pile of mail I just found in the nznog folder, but I thought it might be fun to throw a few observations into the mix: 1. The decision of whether or not to peer with someone always ends up being a business decision, in the absense of regulation, sooner or later. Business decisions are not necessarily driven by the public good. 2. When entities A and B stop peering, performance between A and B usually gets worse (or the cost of exchanging traffic between A and B increases), and usually either A or B wind up paying money to make things better. If A pays the money, then it's a fair bet that the de-peering hurt A more than it hurt B. B wins! 3. New Zealand isn't the first place where this kind of thing has happened, and you can bet it won't be the last. The sky has, as yet, not been observed to fall anywhere else, so predictions of apocalypse may be premature. As an ISP, having your cost structure changed is a pain, especially when you're stuck with 1-2 year contracts for transit that seemed like a good deal yesterday, but which are going to wind up feeling pretty expensive on November 1. Gnashing of teeth and waving of fists at the sky is to be expected. However, aren't TCL just implementing what Global Gateway have been doing all along? Are TCL really getting beaten up over this because they've had a more liberal peering policy in the past? Have people forgotten that the critical mass that has resulted in an APE which is worth connecting to was due (in part, at least) to the original open peering policies of AS 4768, AS 5763 and AS 7714? Did the first peering session across the APE not land at one side in AS 4768? There's a certain angle at which it seems like the more appropriate response is to say "thanks for the peering, it was good while it lasted". (For which, in some small and historic sense, I'm happy to say "you're welcome" :-) Joe