Minutes IETF99: homenet
||Minutes IETF99: homenet
IETF 99 - Homenet Agenda
Monday, July 17th, 2017
13:30-15:30 Afternoon Session I
Grand Hilton Ballroom
0. Administrivia (5m)
Note taker - Tim Wicinski
Jabber relay - Dan York
1. WG Status Update - Chairs (5m)
- draft-ietf-homenet-dot-09 (waiting on updated write-up)
- draft-ietf-homenet-redact-03 (in AD Evaluation, will lapse)
Drafts on hold pending other docs:
- draft-lemon-homenet-naming-architecture (expired)
- draft-ietf-homenet-hybrid-proxy-zeroconf-02 (expired)
- draft-ietf-homenet-front-end-naming-delegation-05 (expired)
- draft-ietf-homenet-naming-architecture-dhc-options-04 (expired)
- draft-ietf-homenet-babel-profile-02 (Juliusz Chroboczek, 10m)
-02 requires a nw sub-protocol for HNCP, and moved to new document.
Committee created, and did not do any protocol work.
Added security section to NOT recommend untrusted links.
update with new reference to source-specific routing; review considerations
Ready for Last Call
- draft-boutier-babel-source-specific (Matthieu Boutier, 10m)
Source Specific extension is required.
Source Specific will require a New format.
Simpler, but incompatible.
3. Naming Architecture and Service Discovery
- Distributed Keys for HNCP, Lemon
HNCP offers key sharing but isn't secure. (DTLS but unclear)
protocol picks a 'shared secret'
* node generate public/private key pair
* node shares public key
* now you can do DTLS
Ted willing to write the spec
Barbara Stark(BS): exactly the right approach, willing to help
Juliusz Chroboczek(JC): Really strong direction
BS: Using X509 self-signed could achieve same goal as DTLS.
TL: Have to elect a signer.
Should have the conversation of certs or keys are the right choice
David Schinazi(DS): this is cool, but one of the sharp edges.
Should be integral part of design
TL: Think I agree with your motivation, but not with putting in this draft.
This draft will go faster.
- Experiences deploying DNCP in home network, Lemon
three HNCP implementations
shncpd - used for readability.
Then things went boom. Stopped working. SAD!
not ready for prime time. we are not done yet.
specifiction has some IPv4 addressing
no naming architecture
integrations that just work.
DM(?) use case different; a bit too much
edge router not running hnetd could not build on Ubuntu
(?): Being forced something to work. some IPv6Ping
Mark Townsley(MT): target market in mind, should be vendors.
George Michaelson (GM): Should not be disregarded. Real Experience.
Lee Howard (LH): Thank you. Homenet hackathon project? no. Interopt Testing
MT: Many OpenWRT options
UI gets graph of top
- draft-tldm-simple-homenet-naming-00 (Ted Lemon, 30m)
Andrew Sullivan (AS): Why host has a theory?
TL: It has to.
AS: Changing the scope to be simpler, which may be the right thing
TL: What would we have done instead?
AS: Always the worry
BS: happy eyeballs to find which connection works better. should be what is best connection
DNS Query different responses based on locations.
happy eyeballs solves DNS
DS: Happy eyeballs solves this
GM: Solved for one dimension....
TL: Looks bad but solves really simple
DM: multi provisioning domain?
BS: No DNS Proxy
Eric Vyncke (EV):
Discovering Provisioning Domain Names and Data
Service Discovery: Discovery Proxy/Advertising Proxy
SC: Talked out.
Every homenet router
using DNSSD to browse zones
JC: Enjoying the discussions very much. Many moving pieces
Sit and think on the list of requirements.
list of scenarios. solve those scenarios.
GM: Right to say close to done. Any confusion at the microphone may be looking
for a much higher abstraction. That is multi-dimensional problems.
Introduce a broker idea.
SC: Thanks Ted for taking initiative and doing this work.
Did not have time to sync up on this. Need Diagrams.
roadmap draft explains how moving parts fit together
AS: Problem getting this published....
TL: Often miss details when not writing code.
Chairs: go to list for CFA
Ray Bellis(RB): Thanked Mark Townsley for his service as co-chair