RE: [v6ops] [EXTERNAL] Re: Scope of Unique Local IPv6 Unicast Addresses (Fwd: New Version Notification for draft-gont-6man-ipv6-ula-scope-00.txt)

["Manfredi (US), Albert E" <[e-mail]>]

From: Ted Lemon <[e-mail]> 

> There are also different uses for ULA. ULA can be used for internal addressing by large orgs, and there there’s potential for overlaps, if for no other reason than that large orgs sometimes merge.
>
> Another use for ULAs is on home networks. In this case, we don’t expect ULAs to ever need to cross the router. So the set of networks on which home network ULAs need to work is very tightly constrained, and we don’t need to worry about ambiguities.
>
> Another use for ULAs is stub networks. In this case, again we do not expect the ULA to ever make it past the adjacent infrastructure link (the link to which the stub network is attached).
>
> So chasing after global uniqueness is not necessary in most cases; even in the case where it is possible that there will be conflicts, /global/ uniqueness is not really the issue. In a case where two orgs are merging, the likelihood of a ULA collision, assuming they used a real RNG to generate the ULA, is small, and if it happens, the worst case scenario is that one or both of the orgs need to renumber before they merge. This is not something that’s going to just randomly cause a problem.

Okay, I agree with those points. Still, to my mind, what is most convincing is how *I*, a user of ULAs, have to go about defining these, and how that would differ from RFC 1918 usage.

RFC 4007 says this much:

   Every IPv6 address other than the unspecified address has a specific
   scope; that is, a topological span within which the address may be
   used as a unique identifier for an interface or set of interfaces.

As an individual net designer, for the fd00::/8 block at least, I cannot allow myself to reuse the ULA prefixes, no matter how isolated my separate nets might be. I have to shoot for uniqueness. Instead, no problem at all using duplicated RFC 1918 addresses, among mutually isolated networks. So, as a designer, this puts an extra burden on me, when using ULAs vs RFC 1918.

And furthermore, for the fc00::/8 block, these ULAs might actually be guaranteed to be globally unique, even if they have to remain inside an admin domain.

Chasing after actual global uniqueness is not necessary, for fd00::/8 ULAs, but it's still what individuals who define these have to shoot for. To me, that's what matters, and it's not very confusing.

Bert