JINMEI Tatuya / 神明達哉 a écrit :
At Wed, 18 Feb 2009 17:56:43 +0100, Alexandru Petrescu
<[email protected]> wrote:
An IPv6 address prefix used for stateless autoconfiguration
[ACONF] of an Ethernet interface must have a length of 64 bits.
I disagree with this. There's an implementation of SLAAC over
Ethernet whose prefix can be shorter than 64 and works ok. I
suppose there's at least another similar implementation.
I don't understand the logic. Simply because there's an
implementation that doesn't follow the standard cannot be a reason
for changing the standard. If we really want to do that, there
should be a clearer and stronger reason.
The pros and cons on the fixed prefix length for stateless
autoconfiguration had been discussed to death, and as far as I
remember the attempt to change the fixed length has never convinced
the community. I'm not necessarily objecting to another attempt by
pointing it out, but I suspect we'll all just waste time by repeating
the same discussion again (ending up no change).
I understand much discussion was about length of the IID.
But this is solely about the prefix, which could be shorter than 64.
Keep same IID length 64.
In this sense - what do you think about an RA containing a 56bit prefix
and used by a Ethernet receiver using 64bit IID?
Out of curiosity, btw, what's the implementation that uses a shorter
prefix (or more essentially in this context, a larger interface
identifier), and for what does it use the larger IFID space?
No, I didn't mean larger than 64bit IID. Same 64bit IID.
Alex
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
