[multipathtcp] Summary for discussions on potential MPTCP proxy work

[Yoshifumi Nishida <[e-mail]>]

Thank you for the replies and discussion about the consensus call and
active discussions on this topic. We apologize for the delay in responding.
We have discussed and studies this topic carefully.



In our observation, we believe most people basically support proxy work in
the WG along with #1 scenario in page 12 at https://www.ietf.org/
proceedings/98/slides/slides-98-mptcp-sessa-chairs-01.pdf



OTOH, we've seen some folks raised concerns on the proposal.



The chairs have discussed on these concerns and we think the following
points will not be solved in the current proposal without significant
updates.



    a) The current proposal requires applications to snoop transit traffics
and modify header information in TCP packets.



    b) The current proposal requires to convey information in TCP SYN
packets. This would need to consider cooperation with other WGs, such as
tcpm or tsvwg, to consider any impact on the robustness and security of TCP.



    c) The current document is hard for people to understand the proposal



Based on our assessment of the current situation, we think some combination
of the followings will be required in order to make progress.


    -   A new draft, based on the existing one but with modifications to
address the concerns

    -   A fresh approach

    -   A possible update of some of our technology criteria. For example,
we've seen some people prefer 0-RTT approach, but this might be a bit hard
requirement and 1 or 2 RTTs could be possible compromise.



Again, we appreciate lots of productive discussions and very sorry for our
slow response.

If you have comments or questions on our current assessments, please let us
know.


Best regards,

--

Yoshi & Phil