Tisdagen, den 15 aug 2006 klockan 19:12, skrev Mark Smith:
On Tue, 15 Aug 2006 11:15:57 +0200 (CEST)
Mattias Webjörn Eriksson <[EMAIL PROTECTED]> wrote:
Otherwise, I conclude that IPv6 is classfull, having two classes:
namely /64 and /128 (routing, though is classless).
I don't think classful is the right way to express this. "classful" is
a loaded word from the IPv4 days, as it defines both an addressing
allocation method and a _forwarding_ algorithm. I think some of the
criticism of the /48 boundary has been based on misunderstanding what
classful addressing was also classful forwarding.
The /48, /64, /128 etc. boundaries in IPv6 are soft boundaries rather
than hard ones, and pretty much there primarily for operational and
administrative convenience. For example, the /64 boundary is basically
there so that an EUI-64 address can be used to generate a
semi-permanent and convenient node address (and also from what I
understand, to allow for routing goop, although we're never going to
"goop").
I still think this is ambiguous (rfc:s might not be, but my interpretation
however is).
To an extent I agree. I think if you look at from a slightly more
operational view point, rather than a technical / protocol one, the
convenience and simplicity of having fixed addressing boundaries makes
more sense.
This sounds resonable. All implementations I've tested so
far happily accept assigning /112 prefix to interfaces.
My gut feeling is that the common practice and implementation is to allow
prefixes between /65 and /127.
Initially I was looking at it from an operational view point, as
I wanted to split a /64 (I had my reasons), but did not know if it was
allowed as per rfc. I found that rfc3513 says that all unicast addresses
(except those that start with binary value 000) requires
an Interface ID to be 64 bit long, thus making /112 invalid.
Current specification says one thing and implementation says another.
Med vänliga hälsningar/Best Regards
Mattias W E--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
