In your previous mail you wrote:
| => it works but I can't qualify the situation of a large part of
| a protocol is only a burden for current uses as OK.
No question but that DHCP is a fairly heavy protocol - but that's at
the server end, and only needed if you actually want to use managed
address assignment. If you do want that on your net, then all of
that baggage is required.
=> my concern is about the managed address assignement: it can be
needed for IPv4 where addresses are rare resources but *not* for
IPv6 where there are 2^64 addresses available per link. You can
believe in address registration for management, etc, but the reuse
of DHCP in the IPv6 world lead to a heavy protocol with a totally
useless main function, so I am *not* happy with this!
Note: I still don't believe that anyone should be required to use
DHCP (with or without the address assignment part) if they don't
want to - but on a net where the administrators want that level of
control, nodes should do the right thing and co-operate.
=> address assignment is not the good way to manage an IPv6 network,
address registration is simpler so better. Unfortunately my efforts
in the past (when I believed DHCPv6 should be used for itself :-)
to add this function in DHCPv6 failed... so we get a protocol
which shall not be fully implemented by major implementors (cf.
KAME and Microsoft) and is already in its 27th version after 7 years!
Regards
[EMAIL PROTECTED]
PS: I wasted in large amount of manpower in DHCPv6 implementations
but I lost my trust in the future of DHCPv6 when it was harshly
rejected at the Redmond IPv6 interim meeting for prefix delegation.
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
