Re: [v6ops] I-D Action: draft-ietf-v6ops-unique-ipv6-prefix-per-host-07.txt

[Mark Smith <[e-mail]>]

On 19 August 2017 at 06:05, Fred Baker <[e-mail]> wrote:
> Hat off...
>
> Why not simply say "a prefix"? We are perpetually wrapped around an axle with the prefix length, but reading this 20 years from now (which is the perspective one should take in writing an RFC - what will it mean when it's history?) I don't see why they would worry about the exact length someone wanted to assign within their own boundaries.
>

I think it matters because prefix length is the compliment of IID
size, and we know that the size and value of an IID has privacy and
security implications.

RFC7421, "Analysis of the 64-bit Boundary in IPv6 Addressing",
suggests that an IID of at least 40 bits is necessary for privacy.

If this draft does not take a position on the size of a prefix given
to a host, it should at least highlight that prefix lengths longer
than /88 can have corresponding IID size related privacy implications,
and may want to reference RFC6973, "Privacy Considerations for
Internet Protocols" and RFC7721, "Security and Privacy Considerations
for IPv6 Address Generation Mechanisms" for more specific IPv6
addressing discussion.

Regards,
Mark.