90 IPv6 sub-TLA allocations made

----- Original Message -----
From: "Sabrina Waschke" <sabrina at ripe.net>
To: "James Aldridge" <jhma at KPNQwest.net>
Cc: <lir-wg at ripe.net>; <ipv6-wg at ripe.net>
Sent: Thursday, August 09, 2001 5:15 PM
Subject: Re: 90 IPv6 sub-TLA allocations made


>
> James,
>
> Your request for a subsequent sub-TLA allocation was evaluated
> according to the Provisional IPv6 Assignment and Allocation
> Policy (4.2.5. Criteria for Subsequent Sub-TLA Allocations).
> Please see:
>
> http://www.ripe.net/ipv6
>
> As you know this policy is under revision and the current
> discussions on these two mailing lists are encouraged.
>
> All three Regional Internet Registries receive input from their
> communities and based on this a new draft of the policy document
> will be presented at the upcoming RIPE 40 meeting in Prague.
>
> The presentation "IPv6 Bootstrap Phase" given at the last RIPE
> meeting in Bologna can be found at:
>
>
http://www.ripe.net/ripe/meetings/archive/ripe-39/presentations/ipv6develop/
>
> A Supernational Registry is a special case. You may want to use
> this presentation as a starting point to discuss the minimum
> allocation size.

Sorry i do not agree, we are not a supernational registry and the only
reason we have not become one is because of the 80% usage rule, but we are
still in the same boat as James. We are not looking at now, and what we need
now, we are trying to be forward thinking and address routing problems we
see now with IPv4 so we do not see them in IPv6. I think the likes of very
large registries should be handled diferently, i do not mean given special
dispensation but looking at the internet as a whole apply what will benefit
the community at large. One thing which will benefit the internet at large
is smaller routing tables especialy since for some time to come IPv4 routes
will have to live next to IPv6 routes. So what i see as a workable solution
is to use a middle structure. Remeber if we do not get it right now then our
errors will be around for as long as IPv6 is.

The middle ground i see as a workable solution is RIR > MIR > LIR. MIR is
multi-national registry which is basicly a group of registries all belonging
to the same AS but with diferant CIDR for their customers. The MIR would be
bound by all RIPE policy apart from the 80% usage. Rather than relying on
the 80% rule which in a supernational registry or a group of registries like
ours is almost impossible to work with aggregation it is based on other
critearia. This critearia could be a solid aggregation plan and projected
usage based on current trends. The MIR would disribute to the LIR based on
aggregation and projected growth. This would mean fewer internal routes and
better summary at the borders and easy filtering. The reason we have not
become a supernational is simply we could not live with the 80% rule
especialy when you look at how fast dial uses address space compared with
others. The LIR could still be responsible to RIPE (though i do not see why
as customers all get /48 or /64) but rather the LIR is responsible to the
MIR but it is the job of the MIR to control the IP addressing in the most
efficiant manner ( i am not talking about conservation). Lets face it
conservation is not a problem with IPv6 and as was shown at RIPE 37 (i
think) will not be for a good 50 years and the sooner we lose this
conservation mind set the better. IPv6 if not approached correctly will be
our undoing, if we have to go with the current structure of LIRs getting
their own chunks we will have internaly now 250,000 routes, mutliply that by
the world and cisco is very happy as their memory sales goes through the
roof. IPv6 have enough oveheads already with neighbot discover and ipsec
built in lets lower the burden or our routers.

Before everyone stands up and starts shouting about doing away with RIPE and
all that implies let me say i am not suggesting this. We do need a regional
governing body and the amount MIR's needed in the RIPE region is not great
and will be the exception rather than the rule. There will still be LIR's
who report directly to RIPE and as has already been stated IPv4 is not dead
yet. The MIR will still have its work cut out justifying its aggregation
policy to RIPE and RIPE will have its work cut out understanding an EMEA
wide routing plan but then thats why we need an MIR.

I will now don my flame proof suit and await the verbal roasting this will
create ;). I await your comments.

Regards,
Stephen Burley
UUNET EMEA Hostmaster





>
>
> Kind regards,
>
> Sabrina Waschke
>
> --
>
> o------------------------------------------o
> | Sabrina Waschke         sabrina at ripe.net |
> | Registration Services Operations Manager |
> |                                          |
> | RIPE NCC             tel +31 20 535 4444 |
> | www.ripe.net         fax +31 20 535 4445 |
> o------------------------------------------o
>
>