[11:37:25] --- geg has joined
[12:10:07] --- geg has left: Disconnected
[13:32:09] --- toro_toro has joined
[13:32:29] --- toro_toro has left
[13:57:15] --- sarolaht has joined
[13:58:33] --- sarolaht has left
[14:20:27] --- mrichardson has joined
[14:24:23] --- momose has joined
[14:24:54] --- KS has joined
[14:33:48] --- okabe has joined
[14:37:11] <mrichardson> i though that there was a scripe?
[14:37:24] <mrichardson> I'm not scribing, I'm taking notes in a window.
[14:42:26] --- toro_toro has joined
[14:43:54] --- KS has left: Logged out
[14:51:19] --- loughney has joined
[14:57:31] <mrichardson> more scribes would help here.
[14:57:46] <mrichardson> seeing as I'm a tourist in this session.
[15:01:12] --- Melinda has joined
[15:13:12] <mrichardson> Erik: sounds like two kinds of BCP.
a) host software creation people.
b) how to configure the network.
Gary: room, should we have two documents?
[15:15:48] --- toro_toro has left
[15:19:53] --- toro_toro has joined
[15:28:35] --- geg has joined
[15:36:07] --- loughney has left
[15:47:58] --- Melinda has left: Disconnected
[15:52:50] --- okabe has left: Replaced by new connection
[15:52:52] --- okabe has joined
[15:52:54] --- okabe has left
[15:55:30] --- geg has left
[15:59:28] --- fp has joined
[16:21:19] --- toro_toro has left
[16:28:32] --- fp has left
[16:34:03] <mrichardson> 15:21:32
Detecting Network Attachment WG (dna)
MONDAY, November 8 at 1530-1730
===============================
CHAIRS: Greg Daley <greg.daley@eng.monash.edu.au>
Pekka Nikander <pekka.nikander@nomadiclab.com>
AGENDA:
5 min Introduction & Agenda bashing
- agenda accepted.
5 min Documents status
Milestone review:
- done, submit to IESG goals for detecting network attachment
in IPv6.
- Dec. 04. submit IESG existing link layer hints catalogue.
adopted as WG draft.
...
Thank you to JinHyeock for draft-ietf-dns-goals-03.txt.
draft-ietf-dns-link-information-00.txt
for review needed.
aim for Dec. 04.
looking at adopting other documents, to be discussed today.
BCP - two of them, would like them merged.
- maybe a design team?
10 min Link information draft
draft-ietf-dna-link-information-00.txt
Eric Nejdjou presenting.
Q: should additional technologies be added?
802.16? 802.3?
there is a scenario where the equipment running the IP layer, it sees
PPP as the link layer, not the GPRS.
Link Up: NCP concludes, etc.
there is some additional situations in 3GGP, where IP runs on top
of the GPRS, without PPP.
Bernard Adoba: a new IP address assignment mechanism for IPv6.
There is one case where there is no PPP, but there is an RS/RA.
speaker: The address context comes from the ADP?
Soliman: the ADP only gives you the link local address?
Discussion about IPv6 assignment in PPP context.
Pekka: Is the link up for the application?
BA: is the address up enough for the application?
the draft is wrong about IPv6 address workings.
Erik: same applies to IPv6 other ethernet.
Greg: take it to the list.
q: could we do .16 hints? commercial requirements?
Ralph Droms: wait until IEEE gets the document in order?
PLEASE DO REVIEWS NOW.
30 min Best Current Practises drafts
draft-narayanan-dna-bcp-01.txt
draft-jinchoi-dna-cpl-01.txt
Retrospect: top-down approach.
- get the structure of the document right first.
Jinchoi: why so much emphasis on bi-directional reachability.
Speaker: if bidirectional can be done, then unidirectional can be done.
if you can have it, then why not?
Erik: If one has 17 routers, do you need to test them all?
Jinchoi: it owuld be good to do bi-directional detection, but
not bi-directional reachability?
Soliman: are we trying to detect movement, or confirm old link
is okay?
Gary: yes.
Discussion about what the current practice for router advertisements?
Pekka: current documents make it hard to actually detect movement.
we need to document what we can actually do now (no changes
to routers)
Pekka: the focus should be what can you do only the hosts.
We also have solutions work for this, where we introduce
new ideas.
Gary: the intention is that we work with any access network, regardless
of whether or not they have the tweaks.
Ralph: I think that this should include advice to network designers on
what not to do.
I.e. have two routers on a wireless network that advertise different
prefixes.
Christian Huitema (CH): please identify the tradeoffs in configuration,
not a straight "do not do X"
Erik: sounds like two kinds of BCP.
a) host software creation people.
b) how to configure the network.
Gary: room, should we have two documents?
Discussion of one document vs two documents.
Discussion of can host methods depend upon network configuration.
Erik: if the documents are split, then one half will not delay the other,
since the review communities are very different.
Pekka: brief explanation of CPL changes.
Erik: discussion of security implications of spoofs of P1 and P2 are
on the same link.
50 min Solutions Work
draft-jinchoi-dna-soln-frame-00.txt
slides.
Pekka: should this document be adopted?
last call, and submit as informational?
Soliman: What if this does not work?
Gary: a) design document published at same time.
b) or have a design document may fail (-> Historical)
There is some issue with interactions with IPv6 WG, and optomistic DAD.
Gary: do not scare IPv6 WG.
Pekka: Q1) do we adopt this document?
Q2) do we proceed quickly with this?
Survey: Q1: 10 hands. to no hands.
Soliman: do not submit it immediately, but do submit it soon.
XX:
Pekka: a framework for thought. a list of approaches.
draft-narayanan-dna-rrd-00.txt
- this is standalone from other things.
- could punt the security to SEND.
Pekka: should it be individual submission or WG doc?
Gary: take it to the list.
draft-daley-dna-det-fastra-01.txt
- see slides.
draft-pentland-mobileip-linkid-03.txt
- create a link identifier to allow hosts to quickly recognize
link changes.
Discussion of using the prefixes as the LinkID?
Pekka: considered using a hash of the prefixes?
Discussion of RA lifetimes vs prefix IDs.
More discussion on the list.
Erik: when you move to a different link, you have to wait longer?
Erik: when moving to a different link, it would be good to get a RA
quickly too.
X: haven't done this yet.
Gary: a failing of det-fastra is that the same router always responds.
maybe you could change which one responds fastest.
maybe the routers could know the sequence.
"precession"
Q: what about minimum delay for preferred router being less than maximum
delay for non-preferred?
X: yes, an issue.
Comparison of solutions chart.
15 min Next steps
Summary:
RA caching draft will proceed as individual draft.
(possible process issue discussed)
Set up a design team for BCP documents to merge them.
room is in favour of this.
DONE: Mon Nov 8 17:35:14 EST 2004
5 min Summary
[16:34:51] --- mrichardson has left
[16:46:01] --- momose has left: Logged out
[22:58:18] --- jishac has joined
[23:00:23] --- jishac has left