From:                  IAB
For Response to:       ITU IPv6 Group
Response requested by: 2011/2/28
In response to:        Liaisons LS 3-E, C 19 - E, LS 5-E, and TD 3  from
                       the IPv6 Working Group
Title:                 CIRs and address allocation
Submission date:       2011/2/25
Contact:               Patrik Faltstrom, paf@cisco.com



The IAB thanks the ITU IPv6 working group for the opportunity to respond
to two liaisons, relating to IPv6 Group contribution C19 "Problems and
Solutions", submitted by the Syrian Arab Republic and the NAv6 study on
the Country Internet Model. The IAB is chartered to provide the Internet
Community guidance and oversight of the Internet Architecture. As such
we have limited our responses to areas in which we believe there are
architectural issues in play.  We have combined our response to these
two liaisons because the issues they address are intertwined.

IPv6 is the next generation IP whose transition is now taking place.  As
requirements for global involvement in IP address allocation developed
the Regional Internet Registry (RIR) system for IPv4 and IPv6 address
assignment policies are developed in the RIR open policy forums. Each of
these provide for governance by all within the Internet Community[1]. 
Allocations are made on a needs-basis, following the goals of
conservation, aggregation and registration.  For issues relating to 
policy allocation, we encourage all interested parties to take part in
the policy development processes that each of the RIRs have put into
place.

The ITU-T has specifically directed the following questions to us:


    i)    Whether this CIR model will affect the Internet routing
          system? and if so, how?

    ii)   whether the introduction of this CIR model will affect
          technically the operation, stability or
          continuity of the Internet? and if so, how?



The IAB and the IETF have studied the routing system extensively.  The
importance of routing aggregation in the current system is documented in
RFCs 4632[2] and 4984[3].  In addition the IRTF's Routing Research Group
(RRG) has spent extensive amounts of time examining the evolution of the
routing system.  We note  the paper delivered to CG2 by the Internet
Society entitled "A Fine Balance".  As that work points out, operational
experience within the Internet technical community has shown that even a
small number of allocation authorities find it difficult to
not fragment the address pool, hence unnecessarily increasing the number
of entries within the global routing table.  The CIR model would
potentially introduce a considerably larger number of registries.

Also, because any model change has inherent risks, in order to
determine whether the change would be appropriate, one must have a clear
understanding of any additional benefits that would be gained through
the change.  To understand that we would need better clarity on what
problems the CIR model is meant to address.  We encourage anyone who has
concerns about the current model to bring those concerns to the RIRs for
consideration, and to make use, if appropriate, of the policy
development process within the RIRs to make appropriate changes to
address those concerns.

We note with appreciation the ITU's interest in furthering adoption of
IPv6, and encourage further collaboration in this area.

The following working groups within the IETF may be of relevance:

grow    Global Routing Operations
v6ops	  IPv6 Operations
6man	  IPv6 Maintenance
rtgwg	  Routing Area Working Group
idr	  Inter-Domain Routing
sidr    Secure Inter-Domain Routing
lisp    Locator/ID Separation Protocol
intarea	Internet Area Working Group

[1] http://www.rfc-editor.org/rfc/rfc2050.txt
[2] http://www.rfc-editor.org/rfc/rfc4632.txt
[3] http://www.rfc-editor.org/rfc/rfc4984.txt
[4] http://www.isoc.org/pubpolpillar/docs/address-allocation_200906.pdf