ULA scope [draft-ietf-6man-rfc3484-revise-05.txt]

[Brian E Carpenter <[e-mail]>]

On 2012-02-15 05:49, Arifumi Matsumoto wrote:
> Dave,
> 
> one quick question below.
> 
> On 2012/02/11, at 11:41, Dave Thaler wrote:

...
>> As such, I believe the correct fix is not to put fc00::/7 into the policy
>> table, but instead to update section 3.1 to say that we map ULAs to
>> multicast site-local scope.   The existing scope rules would then have
>> the effects I argue are the right answers above.
> 
> So, you are suggesting to map ULAs(fc00::/7) to site-local scope also for unicast ?
> Then, I don't see how the former case can be solved by this fix.
> The ULAs should be chosen for both src and dst addresses, if they have smaller
> scope than global addresses, right ?

IMHO we have to be very careful about messing with the formal scope of ULAs.
They are unambiguously defined *architecturally* as having global scope,
and address selection rules have to respect that (unless we make a major
architectural effort, since IMNSHO our whole concept of unicast scope is
broken).

In other words - in the unicast domain, ULAs have to stay global, and longest-
match or explicit rules are the only solution.

Multicast is a different matter, since site-local scope is well defined there.

> 
> Or, are you suggesting to map only ULAs that is in the same prefix to site-local scope ?

Not needed; longest match takes care of that. The tricky case is where you
have several ULA prefixes in use on the same site.

    Brian