[Planetlab-devel] IPv6 support for MyPLC

[maoke]

Marc E. Fiuczynski wrote:
>> technically it is a pretty good idea. but i guess an oui is assigned to
>> a company and the company have full authority to arrange the rest parts
>> in eui-48 or eui-64 identifiers. changing things after their oui without
>> getting their license is anyway not so comfortable. :P :P well, just
>> some feelings. technically i support your idea. :) - maoke
>>     
>
> My proposal does not modify or steal anything away from a company's EUI-48.
> My proposal defines a new EUI/MAC-48 encapsulation into EUI-64.  So unless
> someone else has already done exactly that, I think it is fair game to
> introduce a new EUI-64 encapsulation of EUI/MAC-48 numbers for VM-based
> systems.
>
> Marc
>
>   
the eui/mac-48 encapsulations are said to be the *reserved* format in 
the eui-64 format. it means a company can append their eui/mac-48 
addresses into the eui-64 format; and organizations issuing EUI-64s are 
cautioned against issuing identifiers that could be confused with these 
forms. this may not cause conflict. to my understanding, an OUI is 
assigned to an organization, and then the organization have authority to 
use not only eui-48 but also eui-64 identifiers with its oui prefixed. 
therefore we cannot apply another eui/mac-48 encapsulation format into 
eui-64 space.

in other words, eui-64 is not designed in purpose of moving 48-bit mac 
addresses into ipv6 interface id. in contrast, this is ipv6 who applies 
eui-64 format and follows (and modifies a little) the eui/mac-48 
encapsulations for 48-bit mac addresses.

furthermore, there are really some new mac addresses which are really 
64-bit long, such as firewire (rfc 3146: IPv6 over IEEE 1394), and 
802.15.4 (IETF 6lowpan WG: IPv6 over Low-rate Personal Area Network). 
although our solution mostly concerns ethernet-connected computers, we 
cannot make sure it is not possible that planetlab extend to new sort of 
networks, especially in the age of ipv6. ;-)

anyway, if the oui assignment does mean the company control over either 
24 bits or 40 bits after the oui prefix rather than them both, and we 
are sure ieee 1394 and 802.15.4 and other possible new link-layer 
technologies will excluded from the concern of planetlab, we can do the 
new format of encapsulation.

however, this is my understanding to the eui-64 specification. maybe 
wrong. :P

- maoke