Re: [bmwg] Lars Eggert's Discuss on draft-ietf-bmwg-ngfw-performance-13: (with DISCUSS and COMMENT)

[Lars Eggert <[e-mail]>]

Hi Carsten,

thanks for clarifying!

First, let me point you at https://www.ietf.org/about/groups/iesg/statements/handling-ballot-positions/, which hopefully clarifies the intent behind these IESG reviews and how they are best handled.

Second, let me try and provide a bit more context on the issues you flagged:

On 2022-2-3, at 16:09, Carsten Rossenhoevel <[e-mail]> wrote:
> - "There are a lot of requirements in here that are either no-ops [..], non-sensical [..]"

That was in reference to the TCP paragraph. Specifically:

* "SHOULD use a congestion control algorithm" is superfluous (= a no-op), because the TCP standard already requires that (at MUST level)

* the delay for delayed ACKs is typically specified as a time interval; the document suggests to compute it as a byte size ("10 times the MSS"), which does not immediately make sense

* "Internal timeout SHOULD be dynamically scalable per RFC 793." It is unclear what is meant by the "internal timeout" or how it would dynamically scale. The reference to RFC 793 does not provide clarity here.

> - "This document needs TSV and ART people to help with straightening out a lot of issues"

I was writing this is a hint to the WG on where to go for help with addressing these issues.

> - "The document is also giving unnecessarily detailed behavioral descriptions [..]"

One reason I flagged this is because these details rule out existing (e.g., fast open) or future TCP extensions. Another reason is that they are inaccurate, because although they are detailed they still do not capture all possible permitted behaviors (e.g., connection-close packet exchanges.)

Thanks,
Lars