Matt Zimmerman <mdz@netrail.net> writes: > because you're using THEIR resources to do so, without > explicit permission from them. That's a repetition of the same position that's been stated over and over, without justification. If A sends to B directly in the absence of an advertised route, A is "stealing" resources from B. If B sends to A indirectly through A's transit provider, then B is "stealing" resources from A. What makes the former case worse in your mind than the latter, when it results in higher reliability, lower cost, and a sounder architecture? Reiterating the same position over and over without any basis or logical foundation does nothing to convince anyone that your position is of any merit. -Bill ________________________________________________________________________________ bill woodcock woody@zocalo.net woody@applelink.apple.com user@host.domain.com
Matt Zimmerman <mdz@netrail.net> writes: > because you're using THEIR resources to do so, without > explicit permission from them.
That's a repetition of the same position that's been stated over and over, without justification. If A sends to B directly in the absence of an advertised route, A is "stealing" resources from B. If B sends to A indirectly through A's transit provider, then B is "stealing" resources from A. What makes the former case worse in your mind than the latter, when it results in higher reliability, lower cost, and a sounder architecture?
Think of it this way: Contractual =========== When you peer with someone, you form a contract (whether written or unwritten is for these purposes irrelevant). This says "I will not send you traffic other than to routes that you advertise". Often the legalese behind these in written contracts is amusing, but one thing that is consitent is they use a default route as an example, not as the be all and end all. So if you send traffic to unadvertised destinations, you are in breach of contract. If the ISP concerned wants to avoid a "sounder architecture" (read: "doesn't design his network the way you want him to"), tough luck, you should have agreed something else at the time of the peer. These days peering agreements are business deals. They happen to be business deals that have $0 attached to them for all sorts of reasons Sean D et al explained far more eloquently than I could, but they are still business agreements. Just like any other contract, be in breach at your peril. Which leaves the case where you send traffic to say ISP X *without* any form of peering agreement with ISP X. Contracts with the IXP should prevent this from happening. If you try this, I hope they filter your MAC address. Technical ========= X Y | | ISP A -------------------------- ISP A \ / IXP 1 IXP 2 / \ ISP B ISP C \ / \ / \ISP B-------------------- Let's assume A is a large provider (Sprint/MCI etc) and B is C's transit provider. Let's assume lines between IXP A and IXP B cost a significant amount of money. A may wish only to advertise customers such as Y to C, but not customers such as X as they would otherwise use an expensive resource which C would be paying for. They might wish not to peer with C at all, as they might prefer to interchange traffic with B at IXP 1. But the key point is that as A is paying for the infrastructure, it is up to A to decide how they want it utilised. Sure, you may think it's not the most efficient way to move packets about - if so, go find some capital and set up in competition; good luck - IMHO with open IXP membership the internet would not be scalable without peering policies, and I say this as someone who normally suffers from them rather than enforces them. Alex Bligh Xara Networks
On Mon, 30 Sep 1996, Bill Woodcock wrote:
Matt Zimmerman <mdz@netrail.net> writes: > because you're using THEIR resources to do so, without > explicit permission from them. That's a repetition of the same position that's been stated over and over, without justification. If A sends to B directly in the absence of an advertised route, A is "stealing" resources from B. If B sends to A indirectly through A's transit provider, then B is "stealing" resources from A. What makes the former case worse in your mind than the latter, when it results in higher reliability, lower cost, and a sounder architecture?
In the latter case, there are established agreements for exchange of traffic. In the former case, there are not. B may not even KNOW that A is doing this. The distinction seems rather clear. Besides this, there are engineering reasons why this is a bad idea, many of which have been explained to you already. Also note that dumping your traffic to an NSP at an IXP may not BE a route of "higher reliability" or "sounder architecture". Randomly injecting your traffic into some point on B's network does not guarantee, or even imply, optimal traffic patterns. I also fail to see how this is a lower-cost solution, as, without a peering agreement with B, you must still purchase transit to them from another source.
Reiterating the same position over and over without any basis or logical foundation does nothing to convince anyone that your position is of any merit.
I've seen several messages with excellent engineering, economic and philosopical arguments against this practice. -- // Matt Zimmerman Chief of System Management NetRail, Inc. // mdz@netrail.net sales@netrail.net // (703) 524-4800 [voice] (703) 516-0500 [data] (703) 534-5033 [fax]
participants (3)
-
Alex.Bligh -
Bill Woodcock -
Matt Zimmerman