Minutes IETF98: homenet
minutes-98-homenet-00

IETF 98 - Homenet Agenda

Monday, March 27th, 2017
09:00-11:30 Morning Session I
Zurich D

Chair Slides: <https://www.ietf.org/proceedings/98/slides/slides-98-homenet-chair-slides-00.pdf>

0. Administrivia (5m)

Blue Sheets
Note taker - Barbara Stark, Stuart Cheshire
Jabber relay - Lee Howard

1. WG Status Update - Chairs (5m)

-----

2. Routing

- draft-ietf-homenet-babel-profile (Juliusz Chroboczek, 15m)

09:05 draft-ietf-homenet-babel-profile

Juliusz presented slides:
<https://www.ietf.org/proceedings/98/slides/slides-98-homenet-babel-profile-for-homenet-00.pdf>
(Summary: REQ 5 not being implemented by anyone, what do we do?)

David Schinazi, Apple: recommends removing REQ 5 for the moment.  Can be
restored later.

Pierre Pfister: Should not drop REQ 5. Wishes he could help but can't.
Not sure what to suggest.

Ted Lemon: no sec model, no threat model so no one wants to implement.
Go with DTLS?

Mark Townsley: Asked for clarity on what was said by Ted.

Ted Lemon: We should specify some protocol that we believe will work,
and work out how to do key management/distribution later.

Juliusz:  That is exactly the problem. Babel already has HMAC-based
authentication. The key management/distribution part is what is missing.

Pierre: We can find some way to auto-generate.

David Schinazi:: DTLS and other solutions need a bootstrap. Rushing with
such a solution without figuring out how to bootstrap is not good.

Ted Lemon: A single key is not enough. If we're using symmetric
cryptography, we want a different key for each pair of devices.

Mark Townsley: There are many solutions that have some sort of root key
involved.

Ran Atkinson: OSPF all routers share the same key.  RIP a common key is
used.  No automatic keys in use.  In practice a configuration protocol
is used.

Ray Bellis: Do we have people who would help with item 2 (define
suitable algorithms, define the packet format, provide a reference
implementation)?

Ted: This is not a network operated by someone who manages networks. We
need to do threat analysis and not just jump to a solution.

Small work team to look at issue led by Ted. 5 people raised hands,
including Ted, Pierre, Barbara Stark, Will try to discuss this week.

Lee Howard read comment from jabber room.

----
3. Naming Architecture and Service Discovery

09:28 - draft-tldm-simple-homenet-naming-00 (Ted Lemon, 20m)
Slides: <https://www.ietf.org/proceedings/98/slides/slides-98-homenet-simple-naming-00.pdf>

Stuart Cheshire: Has been leaning more about the Thread Group (See
<http://threadgroup.org>). They are doing good work wrt mesh networks.
There is industry need for this.  One minor correction about discovering
all services on the network: This is in fact possible (for devices that
implement it correctly) by doing a query for "_services._dns-sd._udp".
See "Service Type Enumeration" in
<https://tools.ietf.org/html/rfc6763#section-9>

Kerry Lynn, Verizon: There needs to be a user model articulated before
we do this work.

Andrew Sullivan: It seems like we are throwing out some use cases in
order to simplify.

Ted: Approach I am proposing is very pared down. 

Ralph Droms: ??

Ted Lemon: not considered yet. Would like to hear your thoughts because
you have more experience.

Ray:  Not quite ready for adoption.   Please send feedback to Ted.

----

Ray: Asking Daniel the state of his naming documents.

Daniel Migault: I don't think they are dead.

----

Terry Manderson, reporting on DNSOP review of ".homenet" name:

1. Name could be added to RFC 6761 Special-Use Domain Name registry
2. This could include an insecure delegation from the root zone
3. There is no IETF process for requesting an insecure delegation from the root zone

Wesley Hardaker: What would the secure delegations look like? Would
there be a name server record and what would it point to?

Terry: Yes, there would be a name server record.

Ted Lemon: There are other cases where we have bad stuff? Should we try
to solve more things at this time?

Mark: Ted are you trying to raise the point that it's not just us?

Ted: No. Do what we're doing but have a separate case in dnsops where we
try to solve more. But this is a hard problem and Terry said he doesn't
have an answer to that problem.

Paul Hoffman: Please don't do nothing. If you want something to happen,
consider new working group. dnsops has already failed.

Jari Arkko: Need to be clear on implications of requirements. Make
sensible choices.

Much of this is out of scope of dnsops.

Mark Townsley: We are repeating the discussion from last meeting. We
decided we wanted to try to take the harder route. What is different?

There have been painful discussions.

Andrew Sullivan: The WG is not in charge. This is a IETF document. AD is
asking "Do you really want to go to IETF last call?"

Mark: Consensus of the WG was clear last time. And this time we are
saying the same thing.

Ray: Is there an opportunity to publish in a way where we are not locked
in to the decision?

Terry: Possible. We are in a fluid state.

Ray: It's important that we get a name reserved.

Ralph Droms: Things have changed since last time. There has been dnsops
and other reviews which this WG may want to consider.

Ted Lemon: Is it generally understood that regardless of outcome of
discussion we ned to have discussion with IANA?

Andrew Sullivan: Not as chair of IAB. There is no appetite to re-open
existing agreement. They may ask that we give something up if we re-open
this. Not sure we want the result of what we are asking for.

MOU has effectively been negotiated away from what it actually says.

Stuart Cheshire: The goal was for the namespace to have special
properties and for it to be recorded in a place where those special
properties could be codified.

Terry: More discussion will happen this week. Will send notes to the
list that were used as talking points.