Meeting on IPv6 Address Allocation Policy

[Brian E Carpenter]

Joe,

A multihomed IPv6 site would be expected to have multiple
prefixes, possibly taken from multiple TLAs. But multihoming
is a very hard problem, and we can't afford to delay the start
of IPv6 address assignment in the hope that someone solves it.
What we are talking about is not *whether* subTLA assignment
starts in April, but what the assignment guidelines are *when*
it starts.

    Brian

Joe Abley wrote:
> 
> On Sun, Mar 07, 1999 at 03:01:00PM +0900, Akira Kato wrote:
> >
> > The Regional Registries (ARIN, RIPE/NCC, APNIC) is about to define the
> > IPv6 address allocation policy. Their current draft may be referred
> > from your regional registries. In the case of APNIC, you can get
> > through the URL below:
> >       http://www.apnic.net/ipv6draft.html
> 
> I think that many of the emotional responses that will probably result from
> this draft will centre around the issue that unless a network operates as
> a TLA (or unless sub-TLA advertisements are widely accepted and distributed,
> like the hole- punching that is de regeur in IPv4), multi-homing is a vague
> and complicated issue.
> 
> Given that many organisations which are most definitely not TLA candidates
> have requirements to multi-home (e.g. end user networks), it seems to me
> that a coherent multi-homing strategy is required before regional members
> can comment sensibly on the TLA address allocation policy.
> 
> In the absense of such a multi-homing plan, if I might reasonably say
> "everybody who can show that they are multi-homed act as a TLA". I won't
> say that, because I appreciate the scaling issues involved in the core,
> but if I was an end-user I might.
> 
> Joe