Monday November 8, 2010, 0900-1130, Pearl Conference Room
Chairs: David Bryan, Enrico Marocco (for Brian Rosen)
Note takers: Peter Musgrave and Cullen Jennings, edited/revised by David Bryan.
Status Summary
RELOAD base draft at WGLC (ends today), lots of comments, none deemed substantive.
Diagnostics draft is next in pipeline, has an IPR claim.
Problems with network/webex...change order to wait for base draft presentation until remote speaker can be reached.
Was set aside until RELOAD stabilized. RELOAD is now stable, so moving document forward.
Open Issue #1: Should we have history of decisions in the document?:
Open Issue #2: Do we include app-scenarios in concepts?:
Two updates since last meeting. Draft has gone to WGLC that just ended.
No objections or discussion on minor changes up to end of slide 8, or on slides 9, 10.
Slide 11, Issue: DRR and ICE:
Slide 12, Issue: Node-ID in JOIN/LEAVE:
No objections or discussions on slides 13 or 14.
Slide 15, Issue: Race Condition I
Slide 16, Issue: Join Attach Timing
Ekr: Any objections to timing, or non-editorial issues? (None noted)
David: Has been lots of detailed reviews. No other major issues on list. Does anyone have issues with content or timing? (None noted)
Discussion about DDR, last call issues:
David: Will cut & paste slide recommendations onto email list.
Speaker for DISCO not in the room, present SNMP talk first instead.
Presented stepped through slides
Discussion
David: Has been list discussion, people want use-cases, please iterate doc and discuss on list.
Introductory Discussion
Trust Anchor Slide
David (as chair): Will be a new draft - had discussion here because there was significant list discussion. Please look at this when -01 comes out.
David: PLEASE PLEASE look at Ekr's slides. We want the comments ASAP!
David: Will be more discussion about diagnostics draft. Please look at that one.
David: See you in Prague.
Ended after 1:20
Agenda bash and update David asked people to look at the draft-ietf-p2psip-diagnostics Gonzalo Camarillo asked people to look at IPR on this draft. -------------------------------------- Concepts and Terminology for P2PSIP, David Bryan, draft-ietf-p2psip-concepts-03 1 person had read new version. Issue #1 - On topic of history in the doc ... Go read it and send to the list. Issues #2 Need more input Hope to move to WGLC as information some time soon after reload. ------------------------------------ REsource LOcation And Discovery (RELOAD) Base Protocol, Eric Rescorla (via WebEx), draft-ietf-p2psip-base-11 Two updates since last meeting with new WGLC that just ended. No objections on all the minor changes up to end of slide 8. No objections to stuff on slide 9, 10 Direct Response Routing - Question about if how much of DRR goes in this draft vs the direct response routing draft. - Agreement we need to fix or remove this text. Will discuss with folks doing DRR draft to see which way to go. Node-ID Join/Leave - slide 12 No objects to this. Authors plan to go with this proposal. No objections to proposed changed on slides for: Specifying Counter Values, Pings while Joining Join race condition Bunch of list discussion. Will wait and see what happens on the list before making a decision of what to do. Join Attach timing Slide has sort of been overtaken by the list discussion. Chen and EKR seem in agreement. Plan to go with resolution here. Some discussion about DRR at at end. Will have a new draft out before Dec 10. ------------------------------------------ An SNMP Usage for RELOAD, Wang Wei, draft-peng-p2psip-snmp-00 See slides that explain motivation for this work. Questions at end: David Bryan asked if you still have a central managed SNMP manger. Ans: A few central Question: For the mic: I have several comments here: (1) this seems to significantly overlap with diagnostics. are you suggesting we replace that. Ans: For certain managed network or carriers you have this as well for better monitoring. Q: What about witting to the nodes A: This is for a carrier that owns the network and would be writing to the peers See about 4 people had read the draft. They are not using RELOAD to carry SNMP. There must be something to manage the peers on the P2P network or else they can not be managed. The idea here is to add a module to mange this. Some discussion around if this is in addition to diagnostics or instead of it. Some comments that it would be good to add more use cases. Concern over differences between managed and non managed use case. Conclusion from Chair: Still lots of discussion. Please iterate the draft. Folks please read. Please take to mailing list. --------------------------------------- A RELOAD Usage for Distributed Conference Control (DisCo), Thomas Schmidt, draft-knauf-p2psip-disco-00 There will be a new version of this draft No questions, no comments, no problem. Ended after 1:20 See you in Prague and Thanks
P2PSIP Beijing Mon 9am ==================== Notes from: Peter Musgrave Agenda & Note Well Status Summary - main base draft at WGLC (ends today), lots of comments, none deemed substantive - diagnostics draft is next in pipeline - has an IPR claim Gonazalo: Do we proceed given there is IPR? - has been raised on the list - need feedback on IPR issue and the draft itself (WebEx having issues: skip on to p2p-concepts) p2p-concepts ------------------- [use concepts slides] - was set aside until RELOAD stabilized - RELOAD is now stable Spencer: useful to have, concern that it may be hard to update, want to put in an appendix saying "as of date X" this was the view. Want to avoid discussion about keeping the "thinking" up to date - only read by one person in room - please go read this Open Issue #2 - do we include app-scenarios in concepts? - Cullen (on list) thought scenarios are useful - please examine and express opinion on list - hoping to go WGLC as info shortly after RELOAD RELOAD (Ekr) ------------ - went through slides and the proposed resolutions. Only points which generated any comments are minuted Issue: DRR and ICE David: Are you saying rule DDR only used with no ICE in this doc Ekr: Should we remove flag from this from this doc ? Or perhaps remove the entire section. David: Not sure. Ekr: Let's resolve on list Issue: Node-ID in JOIN/LEAVE - leave in - no objections from the room to proposed resolution Issue: Race Condition 1 David: Still lots of list discussion - can't resolve today Ekr: Any objections to timing, or non-editorial issues? David: Has been lots of detailed reviews. No other major issues on list. Does anyone have issues with content or timing? [no David: I want time to look at DRR Roni: About DRR only works when no NAT and end-to-end connectivity is ok. Can discuss after. Roni: Will there be another WGLC? David: Yes - would give a short period for more comments, since there are a number of changes. BUT please look now and make any issues known. David: Silence implies connect on Ekr's recommended resolutions. Ekr: Flexible about where NoICE/DRR text should go. David: Intent is to figure out where quickly, may punt more to DRR draft. Roni: Should send these recommendations to the list Roni: About DRR, we have a relay mode option...(PM-I missed the rest...) David: Will cut & paste slide recommendations onto email list. SNMP ===== Wang Wei/ZTE [walked through slides] David: Are you thinking of one central SNMP manager for e.g. one central in a carrier deployment David: Is trap info being stored in the Overlay Wei: That has not been determined Ekr: Seems to overlap with diagnostics - are you suggesting we replace that Wei: diagnostics is a basic requirement but operators may want something more sophisticated Ekr: So this is a managed RELOAD ntwk? Wei: Yes, people who want better info Ekr: Should this be read only (don't want SNMP writing to peers?) Wei: Application is carrier, so they might want the ability to write to peers. Ekr: If you have a real managed network - why not just use straight SNMP? Wei: Need magement of P2P network Wolfgang Beck: Not clear what the use cases are here. Can this be made clearer? Spencer: Can we ask who in the room read the draft [four people] Spencer: Is RELOAD carrying SNMP? Wei: No. Ning Jong: No troubleshooting issue - it's monitoring. NJ: Can you give example about why normal user wants to managed by SNMP Wei: e.g. Skype may want to monitor and manage clients - they must have something to do this... NJ: Skype is commercial - we are talking about standards... Wei: It's a choice for operator NJ: Would like to see more scenarios of SNMP usage Spencer: So this is co-operating with diagnostics - not competing? Wei: Yes, diagnostics is basic SNMP is add-on Wolfgang Beck : Need to add more use-cases. Service provider has no way to manage nodes, operator could use this. Roni: Repeat comment about use-cases. You are using RELOAD to get connection from agent-server. If a managed network why do you need RELOAD for this? Roni: We need to see a scenario to see why RELOAD is needed. Wei: We think RELOAD connections are needed since it's different from a traditional mechanism, peers may be behind NATs and cannot use SNMP directly Yu Meng: Have received some requirement, maybe we need super-nodes. Conference system could be a use-case - would need to be well managed. David: See this being useful in "in between" (not fully managed , not fully Ad-Hoc). e.g. centrally managed but implemented in P2P way (e.g. Skype) David: Has been list discussion, people want use-cases, please iterate doc and discuss on list. DISCO ===== [Thomas Schmidt] Ekr: Can someone explain what's new in Disco? Thomas: Speaking about new choices for next rev Trust Anchor Slide Ekr: Why not tie resource ID to public key Thomas: Could use cryptographic IDs Yan: Is question why not use public key encryption Ekr: not clear why URI pattern is better than a key Thomas: agreed - a question of taste perhaps? David: Will be a new draft - had discussion here because there was list discussion. David: Please look at this when -01 comes out David: PLEASE PLEASE look at Ekr's slides. We want the comments ASAP. David: Will be more discussion about diagnostics draft. Please look at that one.