Hi Suresh, On Thu, Sep 22, 2011 at 6:19 PM, Suresh Krishnan <suresh.krish...@ericsson.com> wrote: > Hi Donald, > > On 11-09-21 09:54 PM, Donald Eastlake wrote: >> See below, >> >> On Wed, Sep 21, 2011 at 6:09 PM, Suresh Krishnan >> <suresh.krish...@ericsson.com> wrote: >>> Hi Jari, >>> >>> On 11-09-19 02:35 AM, Jari Arkko wrote: >>>> Following up with a personal comment. >>>> >>>> The draft allocates an interface ID and an EUI-64 MAC identifier from the >>>> IANA block. These are two separate, unrelated allocations. >>>> >>>> The main criticism in RFC 5453 for making additional interface ID >>>> allocations is that old implementations do not know about them and may >>>> collide when making an allocation. I'm wondering if it would be better to >>>> allocate an interface ID that is based on the allocated EUI-64 identifier >>>> per RFC 2464? Then we would at least use the same format as other >>>> interface IDs and a collision would likely mean inappropriate use of the >>>> IANA EUI-64 identifiers. Note that privacy and cryptographic addresses set >>>> the u/l bit to zero, whereas EUI-64 interface IDs usually have it at one. >>>> Sri's draft is silent on what kind of number should be allocated for the >>>> interface ID, perhaps some guidance here would be useful. >>> >>> This sounds like a great idea. I am not sure that IANA has a reserved >>> EUI-64 block like you suggested, but they certainly have a ethernet >>> address block (MAC-48). We can instruct the IANA to assign a MAC address >>> that maps straight into the IID. e.g. >> >> The IANA Considerations for the allocation of EUI-48 and EUI-64 >> addresses are spelled out in RFC 5342. There is no reason for any >> special action to "direct" IANA. > > Not sure about this. There are two registries in play here. One at > > http://www.iana.org/assignments/ethernet-numbers
The registry web page is very messy. One thing I'm supposed to do is to produce an edited version as a suggestion to IANA. Only a couple of small parts of that page relate to IANAN EUI assignments. > and another at > > http://www.iana.org/assignments/ipv6-interface-ids/ipv6-interface-ids.xml The two entries currently in that registry are clearly not derived from IANA allocated EUIs and, in fact, don't look like they are constructed from any EUI. > The first one lists the allocations from the IANA ethernet address block > for unicast use It also lists the more extensive allocations under the IANA OUI for multicast use. > and the second one lists the IIDs to avoid. Even though > there is a one to one correlation between a MAC address allocated from > the IANA block and an IPv6 IID, the second registry is where the > reserved IIDs are listed (as it is possible to have IIDs that are not > based on MAC addresses). It's absurd for these registries to be operated so that they can get out of sync and so that conflicting uses can be allocated for the same EUI / derived-IID. The IANA Considerations for these registries should be updated, if necessary, so that can't happen. I believe that the EUI registry is, in some vague small sense, more fundamental because you can do a variety of different things with an EUI, not just create an IID. Maybe the IID registry should also be updated to show the ranges of IIDs automatically available to EUI holders... Thanks, Donald ============================= Donald E. Eastlake 3rd +1-508-333-2270 (cell) 155 Beaver Street, Milford, MA 01757 USA d3e...@gmail.com > Thanks > Suresh > -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------