Minutes IETF104: lwig
minutes-104-lwig-00

LWIG WG Meeting
IETF 104 - Prague

Room:     Athens/Barcelona
Date:     Tuesday, March 26, 2019, 11:20-12:20
Chairs:   Zhen Cao, Mohit Sethi
AD:       Suresh Krishnan
Presentation materials: https://datatracker.ietf.org/meeting/104/session/lwig
Meetecho for remote participants: https://www.meetecho.com/ietf104/lwig/
Etherpad for notes:
https://etherpad.tools.ietf.org/p/notes-ietf-104-lwig?useMonospaceFont=true

Note Takers: Jiye Park
Jabber Scribe: Francesca Palombini

===============================================================================

1.  Administrative and Agenda Bashing (Chairs, 5 min)
    Note Well, Note Takers, Jabber Scribes, Agenda Bashing

* 5 updated WG documents
* 3 updated non-WG documents

* draft-ietf-lwig-coap-06 expired:
Matthias: no sufficient feedback to move on
Carsten: needs reshuffling with other work, e.g., in CoRE; LWIG chairs can help
by pinging other chairs on this

2.  Carlos: TCP Usage Guidance in the Internet of Things (IoT) (10 min)
    https://tools.ietf.org/html/draft-ietf-lwig-tcp-constrained-node-networks-05

* Numerous feedback received
* Updates: Maximum Segment Size, Explicit Congestion Notification, single-MSS
instead of small-MSS, Delayed ACKs, RTO estimation, number of concurrent
connections, TCP connection lifetime, security considerations, annex (uIP,
RIOT, TinyOS, summary table) * Post cutoff feedback by Stuart Cheshire (more
text on options, MSS < 1200, ) * Ready for WGLC?

Stuart: Good work that helps to counter myth that TCP is to complicated to
implement on constrained devices

Mohit: Please still comment to help shepherd to move forward
Markku Kojo: I promised review but most of my comments covered by Ilpo. Will
provide any additional comments during WGLC.

3.  Rahul: Neighbor Management Policy for 6LoWPAN  (15 min)
    https://tools.ietf.org/html/draft-ietf-lwig-nbr-mgmt-policy-03

* Updates: clarifications and performance results
* Performance test config: LWIP + RPL + neighbor management policy module,
Whitefield framework, UDP data each 10s from nodes to BR * Measure packet
delivery rate (PDR), network convergence time (hard to define, used routing
tables stable for x secs) * Results: >95% PDR, no convergence without
NBR-management (requires large cache size to perform well) * Ready for WG last
call, would like to have more reviews

Mohit: reference for the min priority value?
Rahul: we added in the latest daft

4.  Rene (remote): Alternative Elliptic Curve Representations (10 min)
        https://tools.ietf.org/html/draft-ietf-lwig-curve-representations-01

* Document contains worked-out examples for implementations and specifications
* Updates:– Detailed examples, with formats, for all Curve25519 family members,
expanded security considerations and IANA considerations * Next: document is
ready, but feedback always welcome, need to double-check examples

Mohit: Wait for Stanislav, I promise to review IANA and security considerations
Rene: Should come back before 10 April 2019, probably one week earlier
Mohit: We don't need to wait till next IETF to move forward. Let's wait for
feedback from Stanislav

5. John: Comparison of CoAP Security Protocols (15 min)
   https://tools.ietf.org/html/draft-ietf-lwig-security-protocol-comparison-03

* Updates: Message sizes for key exchanges protocols, reformulations in the
summary of application data * Assumptions for comparison for TLS 1.3, DTLS 1.3
and EDHOC (see slides) * Detailed information to confirm numbers reported * No
numbers for earlier (D)TLS 1.2, only 1.3, but there have been inquiries *
Comparison of message sizes in bytes with/without connection ID * Next: What
does the WG think about the recent changes? Are there other deployment
scenarios to consider? what working group wants to add?

Rene: Compressed TLS document in TLS?
John: It is referenced in next slide, but want to wait for stability. Compact
TLS 1.3 TLS handshake in CBOR potentially will be added

Rene: you have numbers RawPublicKey? if you add Certificate approach, what will
the message sizes will be like? John: The numbers should be easily calculated
from Figure 2 Rene: Would be good to state if protocols have the same or
different security properties in the Security Considerations section. Main
purpose of the document is to show how to squeeze down numbers, but it should
also show if the comparisons make sense John: Will think about this, good
comment Francesca: Conclusion subsection discusses some of this

John: Can see the potential to add Application layer TLS
Mohit: Makes sense to have all the versions of TLS and DTLS. Good to add more
information on why are some protocols more light-weight. For example, explicit
sequence numbers cause overhead, so would be good to document other causes of
bigger message sizes. Mohit: Recommend not to have group OSCORE, simply client
and server. One way of limiting the document is to say that we are not
considering group communication. For the TLS Certificate compression, you could
add it or not, you can decide John: It has been optimized for TLS/DTLS