[arin-ppml] Encouraging IPv6 Transition

John Curran jcurran at arin.net
Wed May 16 16:51:28 EDT 2012


On May 16, 2012, at 2:23 PM, William Herrin wrote:
> I'm just trying to understand what, if anything, you
> think is different about preemptive assignment in this respect.

First, an algorithmic assignment (or preemptive assignment for everyone 
with an IPv4 assignment or an AS number, which effectively will end up
being the same in implementation) becomes a permanently change to the 
IPv6 number resource architecture, i.e. the total space is 100% assigned
once the policy is implemented, even if no one one actually puts their 
individual assignment into productive use.  This is significantly 
different than our existing processes that only assign space based 
on actual requests (and the resulting assignments are far more likely 
to put in actual operational use once made based on actual request)

Secondly, parties receiving assignments often presume that the operator
community will allow the recipient to interconnect to the Internet using 
their provider-independent assignment; as a result, ARIN encourages the 
community to consider the potential routing implications of any address 
issuance policy.  The upper bound on preemptively assigning addresses to 
folks holding AS numbers would (over the long-term) enable an additional 
65K routes, and doing the same for folks holding IPv4 addresses could easily
result in over 2 million unique IPv6 routes (depending on implementation), 
if using preemptively-assigned IPv6 blocks became the dominant model of 
connecting to the Internet.  This is a very different outcome then our 
present IPv6 assignment system, where many organizations make use of 
IPv6 assignments from their service provider due to convenience, and 
therefore connect with no direct increase to the IPv6 routing table.                                                                                                                                     

This does not mean that a preemptive assignment approach is necessarily
good or bad, only that it has different characteristics that the community
should consider in the overall balance of making IPv6 more accessible.

FYI,
/John

John Curran
President and CEO
ARIN


More information about the ARIN-PPML mailing list