They don't actually need a full DHCPv6 server. Support of Information-
request/Reply/Reconfigure would be sufficient.
- Ralph
On Nov 11, 2009, at 1:55 PM 11/11/09, Christopher Morrow wrote:
On Tue, Nov 10, 2009 at 11:13 PM, Ralph Droms <rdr...@cisco.com>
wrote:
My point is that "place with better information" and "best place
from which
to deliver information to the host" are not necessarily (although
likely to
be) the same...
agreed. though 'at my house' and 'at my office' are 2 different ends
of that spectrum (I suspect). So I suspect that 'routers' will soon
have 'mostly full' dhcpv6 servers in them...
-Chris
On Nov 11, 2009, at 12:09 PM 11/11/09, Christopher Morrow wrote:
On Tue, Nov 10, 2009 at 9:34 PM, Ralph Droms <rdr...@cisco.com>
wrote:
OK.
I'll agree that the information about routing changes is
available in the
router. Whether the router has all the information needed and the
mechanisms to translate that routing information into policy
changes for
the
hosts must also be considered.
surely there will be hybrid scenarios where one side
(router/dhcp-server) or the other have 'better' (more)
information. I
think that in the larger scheme, if you provision 'some information'
with dhcp/dhcpv6 you will continue to do that tomorrow.
If there is a network event that triggers host-routing changes you
will have to have coordination between the host & router folks, just
as you do today. If you can have the dhcpv6 server ping it's clients
for an update, so much the better. (even so much better if that
comes
in some 'secure' manner!)
-chris
On Nov 11, 2009, at 10:32 AM 11/11/09, Suresh Krishnan wrote:
Hi Ralph,
I was just commenting that the addressing policy changes
triggered by
routing changes are best initiated by the router and all the other
policy
changes are best initiated by the DHCP server. I was not
commenting on
the
suitability/ease of use of the delivery mechanism(s) at all.
Thanks
Suresh
On 09-11-09 08:58 PM, Ralph Droms wrote:
In the discussion of IPv6 address selection , Dave Thaler asked
me to
comment on this bullet from slide 10:
* DHCP option
- Hard to kick policy reconfigure by a server.
Not wanting to contribute to yet another iteration of the RA-vs-
DHCP
debate, I'm responding through the mailing list. DHCPv6 has an
explicit
mechanism, required by RFC 3315, in which a server can
asynchronously
trigger a DHCPv6 message exchange from the client.
Suresh commented that the router might be a better source of
updates
in
some circumstances, when the selection policy is modified by
changes
in the
routing infrastructure as propagated by routing protocols. I
haven't
thought about that scenario and can't comment...
- Ralph
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------