[ppml] ARIN Board Advises Internet Community on Migrationto IPv6

[michael.dillon at bt.com]

> Because the last 
> 64 bits of the address are required for interface identifiers 
> that only leaves me with 16 bits with which to create a 
> hierarchical enterprise address allocation model. 16 bit of 
> subnetting space is not enough to create subnet allocation 
> model for a large enterprise. 

I think this issue needs to be taken up with the IETF. When the /48
subnet model was first created, it was being put forth as a solution for
a site, not for an entire enterprise. Since then, some people have
applied IPv4 scarcity thinking to IPv6 and decided that a /48 is
sufficient for a single organization. As you quite rightly pointed out,
this overlooks the interface identifier model of IPv6 and perhaps,
recreates a problem which was solved in the IETF years ago. 

You might want to join the IETF discussion list and bring it up there
http://www.ietf.org/maillist.html
However I suspect you may get better info by going directly to the IPv6
working group directly.
http://www.ietf.org/html.charters/ipv6-charter.html
It is best to hurry, because the members of the ipv6 working group are
prepared to wrap up their work next month because, from their point of
view, the work is done. That may be true since deployment of IPv6 really
belongs in v6ops http://www.ietf.org/html.charters/v6ops-charter.html
but the whole division of responsibility is a bit fuzzy to me. If you
feel the same, go straight to the first list above.

>It would seem to me that I 
> would want to use the globally routable /48 for internet 
> facing environments and some form of ULA space internally. We 
> try very hard to employ only RFC based solution in our 
> environments. When are we going to see an end to the debate 
> over ULA-Central and ULA-Local? I have been working with 
> RFC4193 "random" method to create address space for my 
> internal networks but I do not want to move forward with 
> deployment until there is some "ratified" RFC in place that 
> helps to guild my address allocation strategy. To be blunt I 
> don't have time to go through the effort of building a design 
> just to have to rework it because the space the RFC4193 
> defines and ULA-Local gets broken up and allocated to RIRs, 
> LIRs, ISPs, and or carriers to hand out as ULA-Central. I 
> want to adopt v6 and get started but I can't afford false 
> starts and rework. Anyone else of like mind, or opinion? 

Given that ULA addressing is defined in RFC 4193 and that this RFC is on
the Standards Track, I think it is highly unlikely that the IETF would
ever hand that address space over to the RIRs. If you want further
assurances on the status of RFC 4193 then it is best to go straight to
the IETF and ask them on the IETF discussion mailing list.

--Michael Dillon