Opened 10 years ago

Closed 10 years ago

#247 closed protocol defect (fixed)

Scalability questions IANA review for IPv6/IPv4 multicast address request

Reported by: esko.dijk@… Owned by: draft-ietf-core-coap@…
Priority: minor Milestone: post-WGLC-1
Component: coap Version:
Severity: - Keywords:
Cc:

Description

The request for a global "All CoAP Nodes" IPv4 multicast address lead to the following question from the IANA designated expert:

"Have they considered how this may work in sites with Internet multicast connectivity. E.g. there is multicast connectivity between thousands of universities. Do you want a node in one domain to reach all CoAP in all other domains across the Internet?

Does this scale? Are there any security considerations?"

It is not clear from the current CoAP draft how this issue should be addressed.

1) if we follow the guideline that unauthenticated multicast SHOULD NOT be accepted by a server, discovery won't work in typical cases because devices are not mutually authenticated in initial discovery situations.

2) if we state that we won't use this MC address for discovery purposes, we don't need the multicast address registration in the first place.

Change History (6)

comment:1 Changed 10 years ago by esko.dijk@…

Additionally the comment below from the IANA designated expert reviewer, for the IPv6 multicast address request that we made:

"There are some concerns about how "All CoAP Nodes" will be used for IPv4, and it seems there is still discussion in the WG. As the same protocol is used also for IPv6, it may make sense to defer the IPv6 allocation until it is clear how "All CoAP Nodes" will be used for both IPv4 and IPv6, and then process the two applications together.

Is there any need, e.g. implementations and testing, for doing the
IPv6 application prior to the IPv4 application?"

comment:2 Changed 10 years ago by esko.dijk@…

  • Summary changed from Scalability question IANA reviewer for IPv4 multicast address request to Scalability questions IANA review for IPv6/IPv4 multicast address request

comment:3 Changed 10 years ago by zach@…

Coming back to the solution for this ticket. I would like to propose the following in the WG meeting next week:

  1. Change the IPv4 request to the Local Network Control Block. Although we could ask for the Ad-Hoc Block I think Local Network is sufficient for our discovery needs in IPv4.
  2. Change the IPv6 request to link-local and site-local scope only. Here we need site-local scope due to the way 6LoWPAN networks are typically built.
  3. Recommend the use application specific multicast addresses for larger scope multicast needs.

Regarding the security consideration, with these changes I think the existing text in 11.4 is sufficient.

If this is OK with the WG, then we can make the needed changes in -13 and then update our IANA registration request accordingly.

comment:4 Changed 10 years ago by esko.dijk@…

Noting the discussion on the list about an easy-to-remember multicast address similar to BACnet (FF0x::BAC0). Proposals are:

FF0x::C0
FF0x::C0A7

https://www.ietf.org/mail-archive/web/core/current/msg03801.html

comment:5 Changed 10 years ago by esko.dijk@…

  • Priority changed from major to minor

Noting agreement with IANA and address request reviewer:

  • allocation will be done at time of core-coap publication when all details are worked out.
  • IPv4 allocation in Internetwork Control Block is okay; Adhoc Block not needed

comment:6 Changed 10 years ago by cabo@…

  • Resolution set to fixed
  • Status changed from new to closed

Fixed in [1099]:

Close #262.
Close #247.
Close #260 (for now).

Note: See TracTickets for help on using tickets.