[ppml] Policy Proposal 2005-1: Provider Independent IPv6 Assignments for End-sites - Revised Text

[Kevin Loch]

Edward Lewis wrote:

>>            4. be able to assign IPv6 addresses to at least 100,000
>>unique devices within 1 year and advertise that connectivity through
>>it's single aggregated address assignment.
> 
> 
> I'm curious about the rationale for "100,000."  Not that I think it 
> is unreasonable, but was there any reason to pick that amount? 
> Perhaps something to do with the pain of renumbering?

No precision is implied in that number.  The rationale is that lower
magnitudes might allow too many end sites to qualify for PI space.
There is significant concern that policy mistakes made early on will
have long lasting if not permanent effects on routing table size.

100,000 devices appeared to be a number that could overcome
the opposition from the last meeting (as opposed to 10,000 for
example).

Personally I agree with Owen that 100,000 devices is absurdly
high.  I think a good target would be normalizing
with IPv4 multihoming (~400 devices) or non multihoming
(~1600 devices) requirements.

But in the interest of getting the framework in place and
at least some end site assignemnts made I agreed to put
the device requirement in there.  I made sure it was
easy to reduce or eliminate the device requirement in the future
without having to rewrite the rest of the section.

- Kevin