Re: [dc] 答复: Re: Requirement for a method to manage mac address in DC

On 2/3/12 05:45 , Thomas Narten wrote:
> Andy Dockerty <andyd@juniper.net> writes:
> 
>> I am aware that there are concerns about "OUI address space", that
>> not withstanding, the creation of an OUI space analogous with
>> RFC1918 address space, coupled with the extension of OUI
>> registration may offer an option to ensure organizational or locally
>> significant MAC uniqueness....
> 
> RFC 1918 space is shared in the sense that anyone can use it as they
> see fit. Different organizations use the same space, so if you
> merge/join to organizations you risk having collisions. Even within an
> organization, assignments must be done in such a way as to avoid
> collisionss. RFC1918 space is defined to have local scope only, i.e.,
> not be globally unique.

In my experience the by far the number #1 way to produce oui collisions
with VMs by far is when cloning systems or images with predefined oui's.
I'm sort of of the opinion that if you shoot yourself in the foot it
should hurt. attempting to hide or ameliorate that sort of brokenness
would seem like a mistake.

> Seems to me that MAC addrs satisfy that property already, via
> appropriate use of the "local" bit.
> 
> So when you suggest creation of an "OUI space analagous with RFC1918
> space", do we not already have that?
> 
> Thomas
> 
> _______________________________________________
> dc mailing list
> dc@ietf.org
> https://www.ietf.org/mailman/listinfo/dc
> 