[manet] Re: BABEL question

Hi Juliusz,

> -----Original Message-----
> From: Juliusz Chroboczek <[email protected]>
> Sent: Thursday, July 24, 2025 9:25 AM
> To: Templin (US), Fred L <[email protected]>
> Cc: Henning Rogge <[email protected]>; [email protected] List <[email protected]>
> Subject: [EXTERNAL] Re: [manet] Re: BABEL question
> 
> > The point is that RFC4193, Section 3.1 defines a format for ULAs that
> > includes a 40-bit "Global ID" followed by a 16-bit "Subnet ID" field in
> > the /64 prefix.
> 
> Sure.

Right, that makes ULA a "topology-oriented" address range, since portions of
the prefix can appear in RA PIOs/RIOs with prefix length less than /128. I believe
we will want a "topology-independent" address range for which no prefix lengths
less than /128 can appear in RA PIOs/RIOs - what is needed is a prefix from which
only /128 "host routes" can be configured. This is the nature of "MLAs".

> > These values are to be assigned by infrastructure reference points, and
> > not randomly assigned by some isolated MANET router seeking to configure
> > a unique /128.  If MANET routers did that, then there would be
> > opportunity for conflict with a properly-administered ULA prefix if the
> > MANET ever came in contact with the Internet.
> 
> I don't see how that follows.  If you draw a random address in fc00::/7,
> the probability of it conflicting with another node's address is
> infinitesimal, no matter how the other node's address was assigned.
> 
> > while there would be a well-known IPv6 prefix value to differentiate
> > these topology independent "MANET Local Addresses" from
> > topology-dependent ULAs.
> 
> Sure, if the prefix assigned to MANETs is short enough so that collisions
> are vanishingly unlikely.

Right,  RFC4429 is a good reference for understanding collision probabilities.
Still, a "topology-independent" prefix would offer the opportunity of managed
delegation of /128s that are explicitly managed for uniqueness instead of
relying on statistical uniqueness. Imagine a prefix like: "2001:db8:f000::/48"
where the lower 16 bits of the /64 do not represent a subnet ID but instead
identify a host route delegation service. Nodes can then go to the delegation
service to receive a 64-bit interface identifier that is guaranteed to be unique
through administrative assignment instead of relying on statistical uniqueness
properties. The delegation service can then also attest to the fact that a node
is (or is not) authorized to use the /128 address it claims - this is in contrast to
self-assigned random addresses for which there is no attestation service to
confirm their authorized use.

Thank you - Fred

> -- Juliusz

_______________________________________________
manet mailing list -- [email protected]
To unsubscribe send an email to [email protected]