[ppml] Various IPv6 issues

[michael.dillon at bt.com]

> 1) I oppose allowing blanket allocations of /48s by an LIR.
> A /48 is close to infinite for practical purposes but is not infinite.

This is precisely the reason why the IETF's RFC 3177 recommends that
ISPs allocate a /48 to all sites which need more than one subnet. In
IPv4, the limited number of addresses available meant that any network
which needed to grow much more than planned for a year or so, had to do
it by adding new address blocks (additional route table entries) or by
renumbering. Huge route tables, and complex renumbering processes were
both considered to be bad, so the IETF tried to avoid this in IPv6. 

That's why the IPv6 address space is close to infinite, from a global
viewpoint and why a single subnet (/64) is close to infinite from a
local perpective. Therefore, a site is given /48 to try and maintain
this "close to infinite" characteristic at the site level.

> 2) IPv6 renumbering happens.  We need a policy that says a 
> network with X nodes is eligible for a /48 and with Y nodes 
> is eligible for a /32.

In IPv6 we do not count nodes. Instead we count site
allocations/assignments. This used to be /48s but when the /56 was
introduced for residential customer sites, this was modified so that a
normal IPv6 site is counted as the equivalent of 256 private home sites.

>  Traffic engineering is also going 
> to require either A) a new routing protocol, B) changes to 
> BGP, C) deaggregation or D) multiple /32s.

What is wrong with MPLS traffic engineering? 

> I am not sure how ARIN expects these aspects to be handled.
> Guidance here would be useful.

ARIN has never provided guidance on routing.

--Michael Dillon