Re: mboned: Re: IPv6 link-scoped multicast

[Jerome Durand <jdurand@xxxxxxxxxx>]

>>As for the issue #2,
>>our main goal is to *generate* unique multicast addresses - /96
>>automatically (like RFC 3306) without any fear of conflicts.
> 
> 
> "Any fear" is subjective.  It is completely possible that a node 
> might start to send an SSM stream to that particular multicast 
> address.

And remember you can take 1,000,000,000 addresses on the link keeping a 
probability below 1 over 1,000,000 that there is a collision. Therefore 
I think you should justify your proposal with something else (simplicity 
for example, or calculation method to avoid randomization...)

My concern is that we can always derive a multicast address from 
something. You seem to need these addresses and I would really like to 
see examples of applications.

> A constructive suggestion: reword this spec to use the upper bits of 
> FF02::/16, even embedding the EUI64 if you wish, and publish it as an 
> Informational RFC if you think an RFC is really needed.

I'd say FF12::/16 as FF0X::/16 are must be reserved by IANA right?

Thanks !

Jerome


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