Re: [tsvwg] WGLC to publish draft-ietf-tsvwg-multipath-dccp-13 as PS will close 5th March 2024

[Chris Box <[e-mail]>]

Hi everyone

I've read through the draft. It's good work, and the result of a lot of
effort.

I found no substantive issues to discuss, but I did have some thoughts of
an editorial nature. Please find them below. Once these have been
considered (even if they're all rejected!), I would say it's ready to
proceed to the next step.

Thanks,
Chris


1.0 Introduction

Personally I would say the text "When the datagram option is used, MP-QUIC
provides a congestion-controlled unreliable unordered datagram service,
similar to MP-DCCP" is unnecessary. Yes, MP-QUIC is an alternative to this
protocol. But I suspect what readers would find more helpful is a little
more description of the differences between the two. For example, MP-DCCP
is simpler to implement and more efficient for already-encrypted traffic.

One thing it does say is that "MP-DCCP defines procedures that facilitate
subsequent reordering." Why should the reader care about this? I can
imagine a shrug of shoulders. But if you add that reordering has been
observed to result in improved end-user throughput, then they will probably
take notice.

3 Requirements Language

It feels odd this being a top-level section. Normally it's a subsection of
the introduction.

4 MP-DCCP Protocol

Why mention "NN indicates this is non-negotiable" if we are not specifying
NN?

4.1

Typo "ore".

It states that the Version MUST be 0, which appears to conflict with the
following text about negotiating version 1 or 2. If this isn't a conflict,
adding some additional text explaining why would help.

4.2.5

In the first sentence I suggest adding "48 bit", i.e. "The MP_SEQ suboption
is used for end-to-end 48 bit datagram sequence numbers of an MP-DCCP
connection".

4.2.10

This section doesn't explain the relative meaning of different levels from
3 to 15. An example is that if I have a pair of paths with priorities 3,4
will this be used in the same way as a pair with priorities 3,15? It's
probably best for this section to point to section 4.11.

4.7

It would be good to caption the figure to indicate it shows the "most
common state transitions".