Re: I-D ACTION:draft-narten-ipv6-3177bis-48boundary-00.txt

Manfredi, Albert E wrote:
>>A /128 breaks IPv6 Privacy Addresses (RFC3041).  Every device 
>>needs a /64 
>>to allow this mechanism to be used.
>>
>>Bob
> 
> 
> Alternative mechanisms could permit interface IDs to be shorter than 64
> bits, for example 48 bits or far fewer than that. Interface IDs only
> need to be unique within a given subnet, and subnets with long prefixes
> could be small enough to make the uniqueness requirement easy to
> achieve. And also resolve the privacy concern in RFC 3041, by not
> creating interface IDs that are likely to be globally unique?
> 
> If we're quibbling about conserving address space by assigning 56 vs 48
> bit prefixes to individual sites, imagine how much more flexibility and
> savings could be achieved by allowing shorter interface IDs. Without
> imposing the restriction in RFC 3513 Section 4, and without increasing
> privacy concerns. No?

I think this greatly underestimates the practical impact of shortening
the IID. In reality /64 is an architectural boundary, even if in theory
it isn't. I don't believe that revisiting this is realistic. And I don't
believe it is in the least necessary. (This point really concerns
draft-narten-iana-rir-ipv6-considerations-00.txt, not the current draft.)

    Brian

--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------