Re: address portability

[Daniel Karrenberg <Daniel.Karrenberg@ripe.net>]

[note that I am not on most of these lists, apologies iin case 
I repeat what others have said on the lists]

Steve Bellovin has summed it up quite nicely.  The issue is purely a
(technical) matter of routing complexity.  There is a large body of
technical discussion on the matter in the IETF (cidrd and other wgs) and
it is a closed case until somebody invents a novel way of doing IP
routing.  I might add that CIDR and its associated topology driven
address space distribution mechanisms were a 'just in time' development 
as serious routing problems were looming much closer than the horizon. 

Also it is important to realise that number portability is *only* an
issue because of the amount of engineering work associated with
renumbering.  It is not an issue because the numbers are visible to the
end-user and as such have recognition or other value, such as well known
telephone numbers. I am starting to hear less-than-clueful know-it-all
political types call for portable IP numbers 'because we have portable
800 numbers'. That is scary!

A more tenable technical soloution to the renumbering problem that does
not require invention but 'only' implementation is automatic renumbering
which, contrary to popular belief, can be implemented in IPv4 via DHCP. 
It just is not straightorward because doing it right needs a lot of
expertise about its interactions with various related things like DNS and
routing.  It has been done successfully at large sites.  Automatic
renumbering is also an integral part of the IPv6 design. 

Summary: This is a non issue which will be solved by automatic 
renumbering.

Kind regards

Daniel Karrenberg 
RIPE NCC Manager