home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Internet Info 1997 December
/
Internet_Info_CD-ROM_Walnut_Creek_December_1997.iso
/
ietf
/
mobileip
/
mobileip-minutes-96jun.txt
< prev
next >
Wrap
Text File
|
1996-10-07
|
6KB
|
131 lines
Editor's note: These minutes have not been edited
IP Routing for Wireless/Mobile Hosts (mobileip)
Working Group Minutes
36th IETF, Montreal, Canada
(reported by Steven M. Glass <glass@ftp.com>
further edited by Jim Solomon <solomon@comm.mot.com>)
======================================================
================
I) Mobile IPv6, Wednesday, June 26, 1996, 1300 - 1500
1) Agenda Bashing: no items added
2) Administravia:
- All IPv4 docs (except route optimization) have been approved as
Proposed Standard RFC's.
- New co-chair Erik Nordmark <nordmark@eng.sun.com> of Sun
Microsystems announced.
- Routing Area Director Joel Halpern presented the requirements for
advancing the IPv6 mobility draft to PS RFC:
1) Language of the doc must be acceptable to the working group.
2) Need an applicability statement.
3) Normally require implementation (and deployment) but the IESG
is understanding of the limitations on currently using IPv6.
Thus, this requirement will be waived.
3) Discussion on: draft-ietf-mobileip-ipv6-01.txt, Dave Johnson
<dbj@cs.cmu.edu> and Charlie Perkins <charliep@watson.ibm.com>
- Two implementations are under development (IBM & CMU).
- Deregistration is tricky, and the order of events in deregistering
with your home agent should be clearly stated as is done in the
IPv4 draft.
- After much rehashing of old arguments, it was once again decided
that the home agent is indeed a "router" -- i.e. a device which
"forwards packets not addressed to itself". An important
observation is that only routers can receive IPv6 anycast packets,
which has implications to "home agent discovery" by mobile nodes.
- It was decided that Mobile IPv6 need not impose any new
requirements upon the base IPv6 protocol documents. In summary, it
was decided that IPv6 routers SHOULD (as distinctly opposed to
MUST) implement home agent functionality; and, likewise, IPv6
nodes
SHOULD (as distinctly opposed to MUST) implement binding cache
and
insertion of routing headers for routing packets to mobile nodes.
Some key observations about these decisions:
+ Not all routers have static/stable storage (as protection from
hard resets) for storage of binding information for mobile nodes.
A suggestion was made to add a "stable storage flag" to the
neighbor advertisement (or the binding acknowledgement) so that a
home agent could advertise this capability to prospective mobile
nodes.
+ One implication of not requiring ALL routers to be home agents is
that the "anycast" method of home agent-discovery as laid forth
in the current draft is broken. This is because a non-home-agent-
but-a-router-nonetheless could consume the anycast. It was
agreed that a solution to the home agent discovery problem should
be architected and included in the Mobile IPv6 draft.
+ The Binding Update destination option needs to be encoded in such
a way so that nodes that don't understand it send ICMP errors
back to the source. There should also be a return code in the
Binding Acknowledge that says "I support this but I don't have
any room for you in my cache."
+ If a host (that is not a router) supports a binding cache, it
MUST use its current bindings in preference to it's routing table
when sending a packet to a mobile node.
- There was consensus that the Lifetime of a binding should work
similarly to the IPv4 mobility mechanism; i.e. a mobile node
proposes a Lifetime but the home agent MAY reduce the Lifetime to
an amount of time that it is actually willing to provide service.
4) Discussion on draft-teraoka-ipv6-mobility-sub-03.txt, Fumio
Teraoka <tera@csl.sony.co.jp>.
[Although this discussion actually occured in the second meeting
(due to lack of time in the first), it is included here because
of its topic area; namely, IPv6 mobility.]
- Several issues were raised regarding the working group draft. It
is the chair's opinion that Fumio should work with the co-authors
of the working group draft, within the framework set forth in that
document, to address any substantive deficiencies that can be
identified. Topics for discussion included sourcing of multicast
packets by mobile nodes, firewall traversal, and appropriateness of
the utilization of routing headers.
======================================================
================
II) Mobile IPv4, Thursday, June 27, 1996, 1300 - 1500
1) Agenda Bashing: no items added
2) Administravia:
- All IPv4 docs (except route optimization) have been approved as
Proposed Standard RFC's.
- New co-chair Erik Nordmark <nordmark@eng.sun.com> of Sun
Microsystems announced.
- To advance the documents further, requires "significant operational
experience". (cf RFC 1264)
3) Presentations.
A number of presentations were given, each of which has slides that
will go into the IETF proceedings. Several of them focussed on
firewall travesal by mobile nodes, which is seen as a key topic area
to be addressed by the working group in order to provide more broad
applicability of Mobile IP.
Other presentations focussed on fast converging handoff algorithms
and link quality extensions to the Mobile IPv4 registration
protocol.
Finally, route optimization was discussed in which Dave Johnson
summarized the changes to be made in the draft. These basically
boil down to making route optimization work in IPv4 mobility in a
similar fashion to that proposed for IPv6 mobility.