[rrg] ILNP Identifiers

[RJ Atkinson]

On  10 Dec 2008, at 20:46, Xu Xiaohu wrote:
>> % They are not guaranteed to be globally unique
>>
>> Yes for ILNP Identifiers.
>
> Do you mean the ILNP identifier doesn't need to be globally unique,

Yes.  The Identifier is highly likely to be unique,
but it is not guaranteed, and it is not strictly required.

> even statistically globally unique?

I don't know what "statistically globally unique" means.

Something is either "globally unique" or it isn't.  In practice,
forgery means that no Identifier can ever be "globally unique".

(For example, one hopes that a HIP ID is computationally infeasible
to forge without compromising the originating key, but every
deployed cryptographic system I know of does eventually suffer
key compromise, at which point that HIP ID would not be globally
unique either.)

> If so, would you please explain the reason?

Any proposal has to handle the case where some bad party is
deliberately mis-using an Identifier (or Address) that is not
really theirs.  This case is not new to Routing RG, instead
this is exactly like the situation in today's deployed IP Internet
with forged Source IP Addresses. [US CERT: CA 1995-01]

ILNP does handle that case, as I've outlined here before.

Yours,

Ran
rja at extremenetworks.com