Re: [ippm] Mail regarding draft-song-spring-siam

[Haoyu Song <[e-mail]>]

Hi IPPM and SPRING WGs,

Per Alvaro's request, I'd like to initiate the further discussions on the proposals in this draft, especially seeking feedback from IPPM.

https://datatracker.ietf.org/doc/draft-song-spring-siam/05/

In particular, we'd like to discuss two questions:
   1.  new protocol vs reuse existing protocol
   2.  IPv6 EH vs UDP

This draft advocates a general active measurement approach in IPv6 which takes advantage of the source routing feature provided by SRv6. Since SRv6 can pinpoint arbitrary nodes along a planed path, it gives network operators a simple method to collect data for measurements from designated nodes in their networks.    (Path planning algorithms with 100% network coverage have been published by the coauthor and others, which all rely on source routing)

This answers the first question. For the second question: why use UDP rather than IPv6 EH.
(1) UDP is more extensible. There are already other proposals that use UDP to support other measurement protocols. This is a good opportunity to unify the solutions with the same underlying mechanism.
(2) There are issues and limitations on IPv6 EH to be resolved, e.g., size, order, fast path/slow path operation, etc. Using UDP instead of IPv6 EH can avoid such issues and accelerate the adoption and deployment.

We believe active measurement utilizing the latest IOAM is very useful in IPv6/SRv6 for path probing and traffic engineering. Please provide your thoughts on the scheme and suggest how we can progress this work. Thank you very much!

Best regards,
Haoyu


-----Original Message-----
From: Alvaro Retana <[e-mail]>
Sent: Monday, May 22, 2023 11:07 AM
To: [e-mail]
Cc: [e-mail]
Subject: Mail regarding draft-song-spring-siam

Dear authors:

You had sent a message to the WG about the adoption of this document a while back [1] -- the chairs put it in the "adoption queue" [2].

My interpretation of the discussion and the follow-up [3] (cc'ing the ippm WG) is that there was no agreement on the approach or even that a new/different solution is needed.  One of the last messages [4] outlined the points to be considered, but they were not discussed
further:

   1.  new protocol vs reuse existing protocol
   2.  IPv6 EH vs UDP


If you are interested in proceeding, where do we go from here?   Part of the discussion has been whether an SRv6-specific (vs. IPv6-generic) mechanism is needed.  Please settle this question before asking the spring WG to consider the work.  I suggest that you discuss it (in light of specific use cases) on the ippm list (cc spring).


While the discussion is settled, I will remove the document from the WG's "adoption queue".

Thanks!

Alvaro.



[1] https://mailarchive.ietf.org/arch/msg/spring/MoVYcoUw1A9oU56DuVjYIAr6ctk/
[2] https://wiki.ietf.org/group/spring
[3] https://mailarchive.ietf.org/arch/msg/spring/RM6Z3q6kHTRu1V4Zl_dKfUGs-rA/
[4] https://mailarchive.ietf.org/arch/msg/ippm/UufCwJUP0WgiHiYxnG7cp50BOUw/