Thanks to WG Chairs for initiate such an interesting discussion.
I'd like to share my thoughts on this, please see inline...
Original
From: JoelHalpern <j...@joelhalpern.com>
To: SPRING WG List <spring@ietf.org>;
Date: 2023年02月22日 22:38
Subject: [spring] Usage of DoH vs SRH TLVs
_______________________________________________
spring mailing list
spring@ietf.org
https://www.ietf.org/mailman/listinfo/spring
The SPRING WG Chairs have noticed several recent discussions which wound
around to the question of whether specific information belongs in a Destination
Option Header (DOH) before the SHR, or belongs in an SRH TLV. Clearly there
are some pieces of information that are closely tied to the SRH, such as SRH
Authentication information, that belong in the SRH TLV. The question is
discussed here is about information that while tied to the SRH hops is not tied
to the SRH contents.
There seem to be two obvious answers, but we'd like to hear the WG opinion, in
particular to propose alternatives.
One obvious alternative would seem to be simply not to allow any extension in
the SRH that can be properly handled by a DoH, and does not depend upon
information in the SRH other than potentially the current DA, which is in the
IPv6 Destination Address field. This provides a clear decision process for the
working group, but some folks have argued it is limiting or inefficient.
The next obvious choice would seem to be to allow any extension that can be
carried in a DOH to also be carried in an SRH TLV. This seems to lead to a
large number of SRH TLV definitions, complicating implementations and adding
limited value. It should also be noted for this evaluation that RFC 8754
section 4.3.1.1 and section 4.3.1.1.1 make it clear that processing TLVs at an
SRv6 Hop is optional and subject to local configuration.
Does the working group have any suggestions or opinions?
[XM]>>> Yes, generally I prefer the former choice, unless there is explicit
evidence (e.g., implementation and/or deployment) that shows using SRH TLV has
significant advantage.
As to the reasons, except for those already articulated above, it seems there
is a conflict between SRH TLV and SRH PSP if the final destination needs to be
involved, and I'm told SRH TLV processing is not more efficient than DOH.
Best Regards,
Xiao Min
Thank you,
Joel, Jim, and Bruno
_______________________________________________
spring mailing list
spring@ietf.org
https://www.ietf.org/mailman/listinfo/spring