[ippm] Re: A proposal to draft-white-ippm-stamp-ecn

[Greg White <[e-mail]>]

Hi Greg,

Thanks for sharing this.  It does look like an interesting option to consider.  I like that it would support backward compatibility in a way that, I think, makes the SR implementation simpler than the current proposal (and keeps the STAMP message size smaller).  Also it doesn’t pollute the STAMP Sub-TLV space by deprecating a TLV.  This seems cleaner and less prone to confusion for future implementers.

In thinking through the scenarios, it strikes me that it would not be possible for an SR to tell the difference between an “old” CoS TLV and a “new” CoS TLV that sets the REC to 0b00. So, it cannot know whether in this case it is returning a CoS TLV to an SS that understands the new encoding of the RP field or not.   To avoid the possibility of causing backward compatibility issues with existing SS, I think it would require a simple change in the requirements.

Instead of: “Furthermore, such a system MUST add 0b10 to the value of the RP field in the Class of Service TLV in the reflected test packet.”
I think it should be: “Furthermore, if the REC field is non-zero, such a system MUST add 0b10 to the value of the RP field in the Class of Service TLV in the reflected test packet.”

It is a shame that the “ToS/TrafficClass” byte is broken up in a funny way here, and that does result in some complexities for the SS and SR as compared to the proposal in my draft, but in the end that complexity is likely worth it for the other benefits above.

-Greg


From: Greg Mirsky <[e-mail]>
Date: Sunday, March 23, 2025 at 12:28 PM
To: Greg White <[e-mail]>, IETF IPPM WG <[e-mail]>
Subject: A proposal to draft-white-ippm-stamp-ecn

Hi Greg,
thank you for the presentation and discussion. I think that it is possible to address concerns expressed by the TSVWG discussion. I've been thinking about how to add monitoring ECN in the upstream direction of the STAMP test session backward compatible with the specification of the Class of Service TLV in RFC 8972. I put my thoughts in the attached proposal, and greatly appreciate your thoughts.

Regards,
Greg