[arin-ppml] RIPE/ITU

[James Hess]

On Fri, Feb 26, 2010 at 4:15 PM, Milton L Mueller <mueller at syr.edu> wrote:
>> -----Original Message-----
> Note that the ITU proposal for CIRs does not propose to make them exclusive, but rather proposes that an ITU-mediated CIR be an additional option. If one supports competing ISPs, why not alternative address registries? One cannot argue against this option on the grounds that we don't have enough ipv6 addresses to make it viable; we do. One cannot argue against it on the grounds that it messes up the efficiency of routing, because new, RIR-sanctioned NIRs or new RIRs carved out of existing ones would have basically the same effects on routing.
>
As I understand the  NIRs  didn't  receive their own  blocks of
addresses, or get any blocks that were distinctly theirs,
automatically, to do with as they please.   And the NIRs only approve
allocation to users within their region,  the addresses are truly
still allocated by RIR.

The ITU  proposal  implies a breakdown of hierarchical addressing,
which would have profound negative impact on the efficiency of
routing,   if it were to be implemented,  in terms of fragmentation,
number of prefixes, and

Since it then becomes impossible to aggregate at the RIR level and
keep out-of-region routing announcements out of your table, when there
are hundreds of countries applying for and receiving  new additional
allocations  interspersed throughout an ITU assignment.,...


It is not as if the  choice to create  RIRs    instead of  assigning
IP blocks to countries in the first place,  was some arbitrary policy
decision or a   "division of resources"  decision .

Allocating  fixed  sized  blocks to countries  "CIRs" is a polar
opposite to efficient utilization of  IP address space  and
needs-based allocation.


There ought to be some sort of justification criteria that needs to be
met for the creation of new IP address registries,  including respect
for the IPv6 addressing model and hierarchical addressing.

Just like there needs to be justification criteria for the creation of
new LIRs....


Basically it sounds as if  the ITU  wants to  just ignore the IP
addressing model, and get a big  block to do their own thing.     That
 might be fine  if their "alternative"  were to  cleanly   preserve
the hierarchical  properties of IP addressing, and weren't excessively
wasteful.

But it appears that  of the proposal (A) it is excessively wasteful, and
(B)  they really want to break things....


--
-J