There has been considerable discussion about the concept of address ownership and what it means in the context of global routing.
Vince Fuller states that providers advocate using space from the provider's name space because the routing system (as it currently exists) only scales well if the address space is topologically significant.
Andrew Partan says that if this (renumbering effort) only applied to single homed /24, it would have a significant enough positive effect.
Should the InterNIC never give out addresses directly to end-users and should only give it out to ISPs? About 1/3 of the group agree and about 1/3 disagree.
Matt Mathis wonders aloud how we promote the idea of topologically based addresses so that folks get it as a positive thing, not a negative thing.
Bill Norton brings up the available NAT devices. It does exist. It does sort of. Jeff Schiller brings up the point that security protocols will probably not work within this context.
Folks may want to survey the online public documents at the InterNIC.
Sean Doran points how that CIDR has worked, but keeping the route table grown linear is required. Now it is exponential. He also argues that flapping is also a part of the problem and recomputing routing in the face of this change.
Steve Bellovin argues that renumbering is something that we MUST make easily done otherwise legal intervention may result.
Erik Sherk noted that renumbering is not that awful and there have always been transition periods.
Bill Manning mentions that PIER exists to discuss this. He also suggests that folks work towards getting folks to renumber and releasing their historical numbers back to the pool. Someone else suggested that the RA might investigate doing this.
Another person suggested that companies have sales forces that are not well educated about the problem. Another way of seeing this is that some providers do not strongly suggest new customer renumber (if they had a previous allocation).
Vince Fuller pointed out that the original supernet document did strongly suggest that developing a renumbering strategy was required.
Matt Mathis askes about the idea that folks might want to consider settlements based on routing prefixes. 80% would not commit one way or the other.
Someone suggested that part of the problem might be the fact that most everying is using a single vendor product. Competition might help this get done faster. Curtis Villamizar suggested that a routing server would be a good thing to use to help reduce the number of peers in a particular router.
Sean Doran said that convergence is a problem as long as the routing system is dynamic. As more prefixes are added, the reachability and delieverability become reduced.
Walt Prue suggested that everyone pays a tax for routing. The Tax would be waived for folks that choose to renumber.
The Chair of NANOG should send a letter to the Chair of IETF indicating that renumbering tools are a critical path item to the continuing survival of the Internet.