On 30/08/2012 11:00, Havard Eidnes wrote:
Just for the record: I consider his suggestion to be totally outlandish, just the notion that "all one needs at any one time as a user is a single NAT'ed socket" is so far out of touch with what consitutes "need" that the rest isn't really worthy of comments.
That is a polite way of putting it, yes. I agree completely.
However, just for the record, let me say this: the community really should accept that historical assignments should not be put into question
I also agree completely with this.
at least not as part of this discussion, also not at the present time, and it is somewhat doubtful that they ever can be.
The only realistic and straight-forward way to re-claim legacy IPv4 address space is through voluntary surrender by the current holder.
As we're talking about registration in a database, the RIPE NCC has a stewardship responsibility to ensure that the registration details are "correct", for some meaning of the word "correct" which needs to be defined by policy (which is why I was interested in the NREN / NREN member position on the ultimate assignee). As such, the registration policy means that the RIPE NCC will need to perform some level of due process with regard to identifying the ERX holders. This will inevitably lead to the identification of ERX space which is registered to organisations / persons which no longer exist, and in this case there may be a argument to take these addresses out of circulation in order to stop hijacking (whether that means preventing updates or deregistering them/returning them to the IANA pool or something else, I do not know). I would be happy for this not to be part of any initial policy, but I do believe that garbage collection will be a necessary part of the overall ERX management process and that we will eventually need to look at this from a policy point of view. Other than garbage collection, I see no basis for reclaiming v4 address space other than by voluntary return. Nick