+1 If a better alternative is devised for 4rd, as Roland proposes here, then can we deprecate the u/g bit usage? Seems to me that privacy addresses are preferable anyway, if not DHCP or statically assigned ones, so any importance assigned to u/g surely becomes a historical artifact?
Bert > -----Original Message----- > From: [email protected] [mailto:[email protected]] On Behalf Of > Bless, Roland (TM) > Sent: Tuesday, February 05, 2013 12:08 PM > To: [email protected] > Subject: Re: Keeping the 4rd-range issue separate from the general u/g > discussion > > Hi all, > > On 05.02.2013 16:51, Rémi Després wrote: > > Several opinions against a 4rd-range reservation have also been > > expressed, but they concern the purpose of 4rd, or its experimental > > status, or personal feelings against one more specification. They are > > not about technical compatibility with the IPv6 addressing > > architecture. > > > > Keeping the 4rd answer separate from the general u/g bit discussion > > will permit, I hope, to now finalize this answer asap. > > My personal summary is as follows: > - The u/g bits have no real meaning in an IID (though the original > intention of mapping EUI-64/48 addresses as modified EUI-64 into the > IID may have been to preserve their meaning), so using them as > indicator for the presence of a particular "IID format" seems to > be not a good idea (see Brian's draft as well as Randy's "magic bits > are almost never worth the pain"). > > - I feel uncomfortable with reserving another range in the IID space > for 4rd, since it is not clear what implementations should do with > it - building exceptions into implementations (e.g., martian IIDs) > is error prone and may cause conflicts with future IID uses, > even if 4rd is experimental. > So I'm also sharing Ole's concerns here: > http://www.ietf.org/mail-archive/web/ipv6/current/msg16976.html > > - without knowing too many details of the 4rd solution I'd rather > propose to use configured 4rd subnet-IDs as indicator that the > IID has a special format: this would be IMHO far more reliable > and consistent than relying on this "reserved IID range". Since 4rd > will require configuration anyway, one can configure the right 4rd > /64s. I'm aware of the counter arguments (j)(m) in > http://www.ietf.org/mail-archive/web/ipv6/current/msg16949.html, > i.e., "free choice of subnet prefixes at CE sites" and > "IID assignments made before 4rd activation". > But it is not clear to me how often this will be _actually_ the case > in practice and whether it's worth to "sacrifice" IID space and IPv6 > flexibility for keeping these requirements. IPv6 is flexible enough > to automatically reassign subnet prefixes or IIDs, thus I'm still > not buying the arguments yet... > > Regards, > Roland > -------------------------------------------------------------------- > IETF IPv6 working group mailing list > [email protected] > Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 > -------------------------------------------------------------------- -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
