On 10 jul 2006, at 02.21, Iljitsch van Beijnum wrote:
On 7-jul-2006, at 17:14, Kurt Erik Lindqvist wrote:Sean (I think), was the first to start advocating string filtering on RIR allocation boundaries in IPv4. Randy at Verio followed and there where other Tier-1s as well. And life was good.Except for the people who were now filtered with no prior warning and no recourse. The IETF can't and shouldn't want to mandate how people run their networks, but things work a lot better if there is consensus about the parameters within which everyone applies their own judgement.
Problem is that one size does not fit all here. I might want to filter some /48s and others not, depending on who I might think I could push to become a down-stream as well.
Sean at the IEPG meeting in Dec 2000 noted that the only scaleable solution to this is charging per prefix.Actually, this doesn't scale at all. I don't know how many networks that run defaultless there are, and how many of those do so out of necessity, but tracking all of them down and paying all of them some money (how do you decide how much each one gets?) is certainly NOT scalable.
Money in this case will "flow-upwards".
Route filtering is a policy set by the receiving provider. I don't believe there is anything here that the IETF ca say that will make a lasting impression.You may be right, but my reasoning is: the most common IPv4 practice (filter at /24) doesn't translate to IPv6, everyone doing their own thing is suboptimal, and there doesn't seem to be another forum to decide on this, ergo the IETF should step up.
But there is no way to ensure this is followed, actually it's most likely not followed. You can't put 2119 keywords in, so at best you get to document the trade-offs and at the end of the day people will take the approach that will maximise their revenue stream...
- kurtis -