Re: Objection to draft-ietf-6man-rfc4291bis-07.txt

[Fernando Gont <[e-mail]>]

On 03/02/2017 05:57 AM, Lorenzo Colitti wrote:
> On Thu, Mar 2, 2017 at 2:15 PM, Fernando Gont <[e-mail]
> <mailto:[e-mail]>> wrote:
> 
>     among other things, it unnecessarily wastes 18 bits of entropy
> , and> 
> If you ever want that argument to be convincing you'll need to explain
> by how much the additional 18 bits of entropy benefit real users in
> realistic situations.

It's the other way around: if you are randomizing IDs for the sake of
privacy, you should make an argument for wasting 18 bits (other than "I
do it because I can").

You should also make an argument for reusing IDs in different layers and
contexts, something that is known to be a bad practice, and is the
reason for which we have had to work on: RFC7721, RFC7707, RFC7217,
RFC8064, besides patching/updating more than 14 standards.

Besides, since we (IETF) do not specify how link-layer addresses are
generated, it would be very bad to make the security/privacy properties
of the addresses we do specify (IPv6) on the properties of those
(link-layer) that we don't.

That said, if you're keen to waste 18 bits of the address, you should
probably write a draft to propose that the default subnet size/prefix be
82. Otherwise you're literally wasting 18 bits that could be used for
something else.

Thanks,
-- 
Fernando Gont
SI6 Networks
e-mail: [e-mail]
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492