Editor's note: These minutes have not been edited.
Internet Area
Working Group Summary
Montreal IETF Meeting
Submitted by
Jeffrey Burgan and Frank Kastenholz
The following Internet Area working groups and BOFs met at the
Montreal IETF Meeting:
MSGWAY
The messageway working group met for one session. There was a report
on MSU's implementation status. The group heard a proposal on
Multicasting; they were referred to the work being done in ION and
IDMR as possibly having some relevant thoughts. The working group
has also started thinking about security issues. They are facing some of
the same keying and negotiating issues that IPSEC has been dealing
with and were pointed there as possibly having something to offer.
ION
The ion wg met for three sessions. The first dealt with NHRP, SCSP,
and Signaling; the second with IPv6, and the third with Multicast
issues.
The latest NHRP draft was discussed. For all open issues, consensus on
approaches for each were established. The latest revision of the
Classical 2 update was presented. The SCSP draft was discussed; no
significant issues were raised. The draft of UNI 4.0 signaling was
presented and briefly discussed. Further discussions of both SCSP and
signaling will occur on the list.
Three approaches to IPv6 over NBMAs were presented and discussed.
Several major differences between the approaches were resolved by the
group, and any remaining differences will be resolved between the
authors. A merged draft is expected for the next IETF meeting.
The definition of how to coordinate several Multicast Servers within
MARS was accepted as a new work item. An approach to using SCSP
with MARS (primarily an issues list) was discussed. The group offered
feedback and a new draft will be produced. A first draft of the MARS
MIB was presented. A discussionwas held on the topic of scalability of
multicast on NBMAs. The group concluded that this was orthogonal to
the overall utility of NHRP, but that a future work item on the
scalability of multicast was likely.
IP Applications Over Cable Data Network
This was a heavily attended BOF. The intent of the BOF was to
investigate whether there was enough interest to form a working group
to develop specifications for running IP over Cable TV networks. The
First part of the BOF was devoted to three presentations describing the
overall architecture and technology of the cable-tv system to educate
the IETF. The remaining time was occupied by a general discussion.
Several issues were raised as being important for a working group to
address: security, QOS, network management, address allocation,
scaling, and multicast.
IPNG
The IPNG working group met for three sessions.
The first session was devoted to presenting status reports on various
activities related to IPv6. Reports were presented on the following:
UNH Testing; Header Compression; IPv6 Tunneling; Interface IDs; UDP
& TCP/MSS vs. Jumbograms; Neighbor Discovery; ITU Request for
Addresses; Auto-Configuration; DHCP; Mobility; RIP; OSPF; IDRP;
IPv6 over FDDI, NBMA, PPP; Multicast Routing; Routing Table Size
Issues.
The second session dealt with several technical issues. There was a
presentation about Host Anycast which raised several issues, one of
them being whether this is a change to TCP. There was also a
discussion about naming link-local addresses and names. There was
disagreement about whether this functionality was actually needed.
There was also a discussion of how hosts should handle route headers.
The current specification permits hosts to process route headers. The
main issue is should hosts reverse source route when replying to packets
containing route headers. There was also a discussion of BSD API issues.
Several specific functions were discussed and changes to some function
names were proposed. Also, the specification was changed to refer to
the Posix function getaddrinfo(). The router alert option for v6 was also
discussed.
During the third session, there was a discussion about specification
changes and clarifications which resulted from the testing conducted at
UNH. There was also a discussion on multi-homed hosts and the issue
of anycasts. It was agreed that the anycast solution still needs better
definition. There were also discussions about Interface Token Size and
Interface IDs. It was decided that an interim meeting would be useful.
Some of the decisions and conclusions that were reached are:
- Resubmit the "OSI NSAPs and IPv6" to the IESG and request
that the IESG skip the IETF last call and send it directly to the RFC
editor.
- Start a working group last call for the "IPv6 Tunneling"
specification.
- Write a document describing how ISP's should keep IPv4 and
IPv6 routing tables separate inorder to keep the memory required for
routing table managable.
- The default value for the Hop Limit field should be 255.
- It is OK for a router to forward packets with link-local
destination back on the link it came in on.
- The maximum token length for Interface ID's is 48 bits.
PPP
PPP met for two sessions, one for their regular collection of protocol
issue, a second concentrating on the various proposals to do tunneling.
At the first session, the status of the current protocols was reviewed,
including soliciting implementation experience on several of them.
Some minor changes will be made to some drafts and a number of
protocols will then be submitted for advancing up the standards ladder.
A discussion was held about the possibilty of writing an "Experiences in
implementing PPP" document to help folks just starting out. An
announcement was made about the next PPP Connection workshop -- 20
October in San Ramon, CA (contact
[email protected] for more
details).
During the second session, the BACP document was reviewed and
various comments were made. The author will update the draft based
on WG feedback. Also, the issue of tunneling protocols was addressed.
There are currently two protocols which are very similar to tunnel PPP
over IP. There were presentations on PPTP and L2F. The authors have
agreed to combine the two protocols, which will be named L2TP. It is
expected that this will occur in the next several months. There was also
a presentation on a layer 3 tunneling scheme which utilized Mobile IP.
This may form the basis for a new working group on layer 3 tunneling
schemes.
DHCP
DHCP met for two sessions, one concentrating on IPv4 and the second on
V6.
The V4 work was pretty minor. The status of all the documents was
reviewed, including new work such as server-to-server and
authentication (little work has been accomplished on the latter). The
working group will start to think about how to deal with DHCP running
on ATM.
At the V6-specific meeting, a presentation was made on the overall
protocol. A significant discussion ensued on issues related to relay
caching and to getting the same address when you renew your lease. The
caching issue revolves around whether the relays should be able to
cache information that they pass on to nodes; there are many issues
here relating to minimzing traffic on the local link, non-local link,
cache consistancy, and so on. A related issue is to try an minimize the
times when a node must change its address when doing lease renewal
(the goal being to minimise dynamic dns updates and the like).
Service Location
The service location working group met once. A status update was made
to the working group; including a detailed discussion of the changes
made to the draft as a a result of the last call. The group started to
work on some simple service definitions. A presentation was made on
the problems of the interaction between service location, mobile-ip, and
multicast. Implementation status was solicited -- there are 5 known
efforts.
