Minutes for HOMENET at IETF-93
minutes-93-homenet-2

IETF 93 - Homenet
Wednesday, July 22, 2015
1300-1530  Afternoon Session I
Grand Hilton Ballroom
Chairs
Mark Townsley, Ray Bellis

Agenda: as posted at
<https://tools.ietf.org/wg/homenet/agenda?item=agenda-93-homenet.html>

0. Administrivia (10m)
          
	Blue Sheets
	Note taker - TBC
	Jabber relay - TBC
          
1. Draft Status Update - WG Chairs (10m)
          
	Updated WG Drafts:
   	- draft-ietf-homenet-prefix-assignment-07 [in IESG Evaluation (1 DISCUSS)]
   	- draft-ietf-homenet-dncp-07 [in IETF LC]
   	- draft-ietf-homenet-hncp-07 [WGLC closed 14 July, Revised ID Needed]
   	- draft-ietf-homenet-front-end-naming-delegation-03 [WGLC soon]
   	- draft-ietf-homenet-naming-architecture-dhc-options-02 [WGLC soon]
   	- draft-ietf-homenet-hybrid-proxy-zeroconf-00 [no update, waiting on dns-sd]
          
   	New or Updated Individual Submissions:
   	- draft-geng-homenet-mpvd-use-cases-01
   	- draft-jin-homenet-dncp-experience-00
   	- draft-augustin-homenet-dncp-use-case-00
          
2. Selected updates on the above
          
   	xNCP Document Cluster Update (15m)
   	- Steven Barth
          
   	xNCP Implementation Report (10m)
   	- Juliusz Chroboczek
          
   	Experience and Evaluation of the Distributed Node Consensus Protocol (10m)
   	- Kaiwen Jin
          
   	DNCP Use Case in a Distributed Cache System (5m)
   	- Aloÿs Augustin
          
   	Naming Architecture (10m)
   	- Daniel Migault
          
   	MPVDs in Homenet and MIF (10m)
   	- Liang Geng
          
3. Routing (70m)
          
   	Report from Routing Protocol Design Team
   	- Russ White

----------------------
Minutes
----------------------

Administrivia done: Note well, blue sheets passed around, notetaker
named (Barbara Stark), jabber scribe named (James Woodyatt).

-----

Chair Slides
<https://www.ietf.org/proceedings/93/slides/slides-93-homenet-0.pdf>

Chairs read through the Chair Slides. There was no discussion.

-----

xNCP Document Cluster Update
<https://www.ietf.org/proceedings/93/slides/slides-93-homenet-1.pdf>

Steven Barth presented the xNCP document cluster update. Much has
happened.

Mark mentioned that at the hackathon there was an attempt at an
independent HNCP implementation.

Slide 4: There were no objections to any of the DNCP updates.

Slide 7: The team really needs to know if anyone plans on doing a
detailed review and whether they need to wait for that before submitting
to IESG.

Mikael Abrahamsson: What is plan for publication of revision(s) and
working group last call process. Confirming that comments can be
provided after next revision is published.

Mark: Yes, comments can still be provided.

Margaret Cullen (f.k.a. Wassermann): How can this be published in the
absence of routing protocol decision?

Mark: It is distinct. Once there was a dependency. Now there is none.
[Steven and Markus Stenberg concur.]

Steven: Now it just provides PSK and identifies interfaces where routing
protocol can be run.

Margaret: Will read and see if agree.

----

xNCP Implementation Report
<https://www.ietf.org/proceedings/93/slides/slides-93-homenet-2.pdf>

Juliusz Chroboczek presented. Slides are about his experience doing a
2nd implementation of HNCP based on requirements draft. Had minor
protocol comments that Steven/Markus agreed to. Recommends moving
forward.

Alex Petrescu: Question about DHCPv6 and DHCPv4. Must implement?

Juliusz: MUST implement DHCPv4 and MAY implement DHCPv6.

Alex: Will send email to list.

----

Naming Architecture
<https://www.ietf.org/proceedings/93/slides/slides-93-homenet-3.pdf>

Daniel Migault presented. Quick Update. There were no questions or
comments.

----

MPVDs in Homenet and MIF
<https://www.ietf.org/proceedings/93/slides/slides-93-homenet-4.pdf>

Liang Geng presented. This was also presented in mif.

Teco Boot: If each router sends an RA with all this info why do we need
this also as a way to send out the info?

Mark: Because there is no "host" vs. "router" side of the box, and HNCP
is our info flooding mechanism, we think PvD info needs to be sent out
via HNCP.

Suresh Krishnan: Explained that what's important is for the integrity of
the information to be maintained. The problem is that there may be boxes
between the CE router and the HNCP box that don't know PvD and this
breaks PvD integrity.

Margaret (as mif chair): It's clear there are some things wrong with PvD
DHCP option as it relates to homenet. We need to get issues resolved
quickly. mif is holding up progressing their doc until this can be
resolved.

Terry Manderson (as AD): Get work done fast and come to an agreement.

Brian Haberman (also an AD): DHCP option inter-relationship needs to be
understood.

Bernie Volz: DHCP doesn't have a problem, but we all do need to
understand the issues from having both.

Mikael: There's nothing we need to do for DHCP from our end.

----

DNCP Use Case in a Distributed Cache System
<https://www.ietf.org/proceedings/93/slides/slides-93-homenet-5.pdf>

Aloÿs Augustin presented:

Mikael: What was the rational for implementing new instance of "DNCP"
instead of using all of HNCP?

Aloÿs: Didn't need much of what's in HNCP for this.

Mark: It's all my fault. I asked Aloÿs to do proof of concept for just
this function.

----

Experience and Evaluation of the Distributed Node Consensus Protocol
<https://www.ietf.org/proceedings/93/slides/slides-93-homenet-6.pdf>

Kaiwen Jin presented.

Mikael: Thank you for doing this. There was no info in the TLVs? It
would be interesting to do tests again with populated TLVs.

Kaiwen: Correct. Agree that it would be good to do tests with TLVs
populated.

Juliusz: Graphs on slide 8 suggest it is trying to converge too fast and
it may be good to delay some messages. OSPF and IS-IS have worked on
this problem extensively.

----

Report from Routing Protocol Design Team
<https://www.ietf.org/proceedings/93/slides/slides-93-homenet-7.pdf>

Russ White presented.

Slide 7: Note that "IS-IS more likely to be understood and supported by
providers offering managed services in the home" was not in the original
charter.

Barbara Stark: (1) homenet by definition is an unmanaged network so
service provider desire to manage a home network belongs in anima, not
homenet. (2) The people who do home network support are very distinct
from people who do network or enterprise support and know nothing about
IS-IS.

Wes George: Not about turning homenet into managed network but about
providing ISP with tools they are used to that can be used from ISP
router to help users.

Slide 12: Traffic engineering requirement is unclear.

Juliusz: I wrote 4 page draft explaining what was needed. Babel does
this.

<additional discussion about metrics, etc. from David, Steven, etc.)

Ray: Stop. Not going to discuss which protocol does or doesn't do
something. The requirement was unclear and needs to be clarified.

Slide 13: Internal realms requirement is unclear.

Slide 20: There is no winner and no recommendation.

Lorenzo: What about the task of estimating time it would take to make
candidate protocol fully viable?

Russ: AD told us not to do that.

<unproductive discussion occurred>

Barbara: Design team work was valuable. It shows both candidates are
viable. Pick one.

Ray: Clarify remaining requirements on list, quickly.

Russ: Would prefer draft contribution to document clarifications so
there is something more easily pointed to.

Steven: Not ready to make coin toss.

Terry/Brian: <Described process to get decision when group is
deadlocked.>

Dave Taht: The fact that people aren't actually trying the proposed
solutions in their own home networks frustrates me. OpenWRT loads of
both exist.

<discussion occurred that did not seem to introduce new ideas>

Steven: Need to separate HNCP and the routing protocol.

Margaret: People don't wait for protocol to be standard to implement.
You can pick without everything being published. But there does have to
be one "mandatory to implement".

Andrew Sullivan: Somebody should just pick one.

Alia Atlas: Let Russ finish.

Ole Trøan: Recommend not picking and finishing other work.

Lorenzo Colitti: Is there an IETF problem with open source code? Could
someone just bring in a proposal as an individual submission?

James Woodyatt (as Jabber scribe): People on Jabber say there just needs
to be a recommendation.

John Brzozowski: If IS-IS is selected, it will be changed to meet the
needs of the homenet environment.

Alia: If Babel wants to be a standard, people need to come to IETF and
do it.

Mark: Open Source community likes Babel and thinks IS-IS needs effort to
work "here" (in the world of Open Source).

Ray: In closing, next step is to create draft that clarifies unclear
requirements (as identified in slide deck).