 |
 |
 |
 |
|
 |
|
 |
|
 |
|
 |
|
|
|
|
|
|
|
|
|
|
|
|
Brian E Carpenter wrote:
This creates a black hole not only for Site A, but also any other sites in the Internet, if ULA is visible in DNS. So, preventing a black hole only for ULA site does not suffice.
True. Which shows that our scope model is too simple for the real world.
While writing my suggestion on a new address selection algorithm, I've tried to tackle this problem too. How can a machine react to different non-link-local scopes correctly if the machine can't even hope to identify them?
The best I've come up with so far is as follows (and I think it could be applicable to any other rfc3484 revision):
If the autoconfiguration mechanism supplies a default route for the address, it is a global scope address. Otherwise it is only site or link local.
Same goes for static configuration, although it will probably be harder for the machine to recognize which address a staticly configured default route is for in the general case...
RFC3442 and RFC4191 allow autoconfiguring more specific routes too. Site scoped addresses should be supplied with routes to the address space they are usable with. Obviously we'd need an IPv6 equivalent of RFC3442.
(DISCLAIMER:)How or whether the address selection algorithm uses these more specific routes is IMO less relevant than the fact that it could recognize address scope from their presence or absence...
-- Aleksi Suhonen Department of Communications Engineering Tampere University of Technology