Re: draft-gont-6man-managing-privacy-extensions-00.txt

[Brian Haley <[e-mail]>]

On 03/12/2011 06:29 AM, Fernando Gont wrote:
> On 09/03/2011 08:17 a.m., Mikael Abrahamsson wrote:
> 
>>> I recommend that folks read the above draft.  I haven't seen the
>>> I-D announcement get cross-posted to the IPv6 WG, perhaps due to
>>> the volume of recent I-D postings, and the topic seems relevant.
>>
>> I don't think it solves what it thinks it solves, but if this REALLY
>> should be implemented, it's my initial thinking that the H flag should
>> be a MUST demand to only have ONE and only one MAC-based IPv6 address
>> according to EUI64. I would appreciate some reasoning in the draft why
>> this was chosen as a SHOULD option.
> 
> I realize I misunderstood your comment.
> 
> The intention of the I-D is/was that if the "H" bit is set, you SHOULD
> generate your addresses corresponding to this prefix ONLY with according
> to EUI64.
> 
> The above requirement is a "SHOULD" (rather than a MUST), since a hosts
> that has good reasons to do so (e.g., privacy concerns) is allowed to
> override the policy specified by the router.

I have an almost off-topic comment, but since I've seen no mention of it
in any of these privacy threads...

You have to assume in a large data center that almost every MAC address you
encounter is going to be randomly generated.  This is because there are
*lots* of virtual machines out there, and the density of them is increasing
exponentially.  Their MACs could change daily as they are deployed and
decommissioned, but from their perspective their MAC is hardware-based.
I know you have this as a "SHOULD" requirement, but figured I should mention
it since it's one of those places where this bit will be completely ignored.

-Brian