Le 13/05/2011 15:37, Thomas Narten a écrit :
Per a previous thread, there are indications that the WG may now be
willing to recommend that DHCPv6 be a SHOULD for all hosts. This is
based on the following rationale:
Thomas Narten<nar...@us.ibm.com> writes:
I personally would support having DHCP be a SHOULD rather than a
MAY. The justification in my mind is that if you want the network
operator to have the choice of whether they want to use Stateless
addrconf OR DHCP, they only have that choice of devices widely
implement both.
This was supported by some others, particularly now that it is clear
there are more implementations of DHCPv6, e.g.:
Bob Hinden<bob.hin...@gmail.com> writes:
While my personal view is that DHCPv6 won't be used for host
configuration in cable/DSL deployments (except for provisioning the
prefix to the home router), it appears that DHCPv6 is being widely
implemented in host OS's because it is needed some environments.
There are enough variations in deployment models that a host
developer will need to support both.
Based on this, I think a SHOULD is OK.
Let me propose the following change be made to the node requirements
document:
OLD/Current:
DHCP can be used to obtain and configure addresses. In general, a
network may provide for the configuration of addresses through Router
Advertisements, DHCP or both. At the present time, the configuration
of addresses via stateless autoconfiguration is more widely
implemented in hosts than address configuration via DHCP. However,
some environments may require the use of DHCP and may not support the
configuration of addresses via RAs. Implementations should be aware
of what operating environment their devices will be deployed. Hosts
MAY implement address configuration via DHCP.
New:
<t> DHCPv6<xref target='RFC3315' /> can be used to obtain and
configure addresses. In general, a network may provide for the
configuration of addresses through Router Advertisements,
DHCPv6 or both. Some operators have indicated that they do
not intend to support stateless address autoconfiguration on
their networks and will require all address assignments be
made through DHCPv6. On such networks, devices that support
only stateless address autoconfiguration will be unable to
automatically configure addresses. Consequently all hosts
SHOULD implement address configuration via DHCP.</t>
YEs I support this text as a SHOULD DHCP.
I have several positive reasons for that, related to efficient
auto-configuration in vehicular environments, moving networks using
prefix delegation - better use just one DHCP, instead of mixed DHCP+RA, etc.
But there are also unknowns: we still dont know how to configure a
default route using DHCPv6. One issue is the lifetime field (Default
Router List structure of ND implementation, and RA, use 16bit lifetimes
for the Default Router, MIF-DHCP draft doesnt say how long a lifetime
is, and DHCP lifetime fields are typically 32bit). These aspects may be
under discussion but until solved we dont know. (I know here locally,
but may be different at yours). There are other issues.
Alex
Is this acceptable?
Please respond yes or no. Given the WG's previous hesitation to having
DHCPv6 be a SHOULD, it is important that we get a clear indication of
whether or not the WG supports this change.
Thomas
--------------------------------------------------------------------
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
--------------------------------------------------------------------