[6bone] proposal for transfer of 6bone address management responsibilities to RIRs

[Roger Jorgensen]

At 08:42 PM 8/20/2002 -0300, Rik van Riel wrote:
>On 21 Aug 2002, Robert Kiessling wrote:
> > Rik van Riel <riel@conectiva.com.br> writes:
> >
> > > Agreed.  6bone is essential for those areas of the internet
> > > where the "normal" providers do not offer ipv6 yet.
> >
> > 6bone is a major source of instability for the production IPv6 network
> > and thus hinders deployment of IPv6.
>
>[snip]
>
> > Well-controlled end sites and address allocation to them for
> > experimental purposes is a different story, but them must be isolated
> > so they cannot wreak havoc to the rest of the world.
>
>Agreed.  We want to isolate the production and experimental
>ipv6 networks somewhat, to make sure the production network
>stays stable and won't be affected by instabilities in the
>exerimental network.
>
>I just hope this will be done in a way that doesn't cut any
>of the current ipv6 users from the ipv6 internet.

If we can assume that 6bone will not be used for production site
any provider that don't want to have a connection to 6bone can put
filter on who they allow 3ffe::/16 from, possible remove them. It
allow ISP's to decide if they want to provide access to 6bone or not.

Of course we can't assume that all production sites only will be
runned of RIR space so the solution isn't good, but it might be
one way to divide 6bone a bit from the rest.

But, seperation of 6bone from RIR will not garanty stability in the
network, let's use the recent AS1654 episode as example again,
how can we avoid that one singel AS announce all sTLA?

In IPv4 you filter very strict on what you allow and what you aggregate,
is this one way to go for IPv6?  Or should we try to come up with
another solution to avoid one singel AS to announce ALL sTLA's?
(no AS should announce more than one sTLA normaly...)




---
Roger Jorgensen (rjorgensen@chello.com)
System Engineer @ UPC Technology / IP engineering
handles: ROJO1-6BONE ROJO9-RIPE RJC10-NORID