[ppml] NANOG IPv4 Exhaustion BoFRe: NANOG IPv4 Exhaustion BoF

[John Curran]

At 4:19 AM -0800 3/7/08, Randy Bush wrote:
>at layer 11, i am hearing a lot of noise on the line of "registries
>desperately trying to assure a continuing income stream,"

Registries have a function to perform (technical administration
of numbering resources) which right now doesn't appear to be
readily automatable.  This is mostly because the semantics on
the resources include both uniqueness and hierarchy/topology,
with the result being historically lots of wrangling over allocation
policy and routing implications.  As IPv4 availability decreases,
the expectation appears to be even more policy work to 'best'
administration the remaining free pool.

If the semantics were changed to be only uniqueness (such
as you might have in the IPv6 domain after any successful
endpoint/location split, or IPv4 post-depletion if it's felt that
the routing implications of IPv4 xfers are simply beyond RIR
scope), this core function of the RIR becomes much simpler
and one can imagine a greatly reduced need for policy
administration and a high degree of automation on the
resource administration side.

When considering implications of a potential 'market', one of
the points raised by the ARIN AC to the ARIN Board was best
summed up as follows:

  *  Preserving ARIN's reason to exist, should not enter the
      calculus of this decision.  If the path to a better Internet
      involves the dissolution of ARIN, then so be it.

I don't know about the thoughts of other RIR's on this topic
(and the ARIN Board hasn't expressed a formal position) but
personally I feel that the RIR's exist to serve the need of
technical administration of the resources, and most definitely
not the other way around.

/John