Early Review of draft-ietf-6lo-nfc-10

Request Review of draft-ietf-6lo-nfc
Requested rev. no specific revision (document currently at 13)
Type Early Review
Team Internet of Things Directorate (iotdir)
Deadline 2018-09-28
Requested 2018-09-05
Requested by Suresh Krishnan
Other Reviews Intdir Early review of -10 by Sheng Jiang (diff)
Tsvart Last Call review of -12 by Wesley Eddy (diff)
Opsdir Last Call review of -12 by Qin Wu (diff)
Secdir Last Call review of -13 by Leif Johansson
Genart Last Call review of -12 by Jari Arkko (diff)
Review State Completed
Reviewer Brian Haberman
Review review-ietf-6lo-nfc-10-iotdir-early-haberman-2018-09-25
Posted at https://mailarchive.ietf.org/arch/msg/Iot-dir/vUNUS1LadBoZm0yun2bXUhYSbCY
Reviewed rev. 10 (document currently at 13)
Review result On the Right Track
Draft last updated 2018-09-25
Review completed: 2018-09-25


* Section 3.4 mentions the use of both UI PDUs and I PDUs to carry IPv6 packets. However, Section 3.2 only mentions the use of I PDUs for carrying IPv6 packets. This seems to be a discrepancy that needs to be cleared up.

* Section 3.4 has the formula MIU = 128 + MIUX. Should MIU be MTU? Additionally, the use of MIUX is only stated as a SHOULD. If the MIU is 128 and the minimum IPv6 MTU is 1280, shouldn't the use of MIUX be a MUST? The text in Section 4.8 seems to mandate the use of MIUX (i.e., "MUST be configured with an equivalent MIU size to fit the MTU of IPv6 packet"). Given that Section 3.2 says that the NFC LLCP does not fragment/reassemble, I am not sure how NFC can support a minimum IPv6 MTU without MIUX. If it can, can you specify a scenario where the lack of the MIUX is not a problem?

* Section 4.3 says to use RFC 7217 for generating IIDs. Given the small size of the NFC address space, should there be some guidance on how the Network_ID parameter of f() can be used to increase the randomness of the generated IID?

* Section 4.3 refers to RFC 7136, which says that the 'u' bit is meaningless unless the IID comes from an IEEE MAC address. However, the text then says that the 'u' bit of the generated IID MUST be set to 0. Why?

* Section 4.5 seems incomplete. Is IPv6-over-NFC only viable in the mesh-under topology? Are there 6LBR <--> 6LBR scenarios in NFC that need to be described?

* I don't see any technical benefits in Section 5 as it does not specify any interoperability details.

* I don't see where anything from RFC 4944 is used in this document. The only text that refers to 4944 says to not use the FAR procedure from 4944. Either the reference should be removed or this document should specify which parts of 4944 should be applied.