home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Internet Info 1997 December
/
Internet_Info_CD-ROM_Walnut_Creek_December_1997.iso
/
ietf
/
dhc
/
dhc-minutes-95apr.txt
< prev
next >
Wrap
Text File
|
1995-05-26
|
4KB
|
96 lines
CURRENT_MEETING_REPORT_
Reported by Ralph Droms/Bucknell University
Minutes of the Dynamic Host Configuration Working Group (DHC)
The Current Mobile IP Protocol
Charlie Perkins gave a presentation about the current Mobile IP
protocol, and about the use of DHCP in giving HA (Home Agent) addresses
to Mobile IP hosts. This use of DHCP will require the use of a new DHCP
option. Charlie has contacted IANA, which has allocated DHCP Option 68
for HA address distribution. Charlie will publish an RFC describing the
use of Option 68; the working group suggested a separate RFC as the use
of Option 68 is more complicated than the use of the other DHCP options.
Two New Messages
Ralph Droms spoke to the group about two new messages, DHCPREVALIDATE
and DHCPINFORM, that were proposed in San Jose. Weak response from the
working group indicated that DHCPREVALIDATE was a bad idea (and will be
deleted from the protocol specification) and DHCPINFORM was a good idea.
DHCPv6
Jim Bound talked to the working group about DHCPv6. The majority of the
group had not read the IPv6 specification which resulted in some
questions due to lack of familiarity with IPv6. The group discussed the
following issues:
o What opaque identifier should be used for the Interface Token (as
defined in the IPv6 stateless autoconfig specification)?
o Transaction ID - is it really needed for the request/response
protocol, and what are the precise semantics of its use? The
consensus was that the transaction ID will be needed and the
details of its definition will be included in the DHCPv6
specification.
o The ``addrs available'' field was considered, as currently
specified, to be of uncertain utility as the number of available
addresses may change between client queries.
o Bit field versus integer values as specifying the contents of a
particular message. The consensus was that bit fields would meet
the technical requirements.
o The request/response model was reviewed and no objections were
raised.
Further discussion of the DHCPv6 specification will take place in the
mailing list, to allow, if possible, completion of the specification
prior to the Stockholm IETF.
Greg Minshall's Model
Greg Minshall gave an encore description of his model for server-server
interaction in support of reliable operation of replicated DHCP servers.
There was sufficient interest to warrant the formation of a new mailing
list; the list will be set up and announced to the host-conf mailing
list.
Client and Server Authentication in DHCP
Laird McCulloch made a pitch for client and server authentication in
DHCP. There was some lively discussion about the utility of such
authentication; any malicious host can simply guess or deduce the
parameters handed out by DHCP, so denying DHCP parameters to a
non-authenticated client gives no additional protection against those
malicious hosts. However, after some discussion of the merits of not
introducing any additional security holes through DHCP, protection
against accidental initialization of unwanted hosts, and the need for
protection against spoofing servers, the working group agreed that an
authentication mechanism would be of some value. The consensus was that
a three-way private key exchange, perhaps CHAP from PPP, would be easy
to add to DHCP through the use of option fields. A discussion of this
authentication scheme will take place on the host-conf mailing list.
Items From The Floor
There was discussion of ``freely available'' DHCP implementations --
there is an implementation from the WIDE project in Japan, and hints
that a freely available implementation may be considered by another
group -- and an expression of interest in a bake-off in the San
Francisco area some time in June.