Re: [Detnet] WG Last Call: draft-ietf-detnet-bounded-latency-02

[Mohammadpour Ehsan <[e-mail]>]

Dear Toerless,

Thanks for your comments. A new version is attached to this email that addresses your concerns. More specifically,

Alas, i do not see the concerns i raised with the authors about section 6.5 solved
that i send before IEF 110 in email. If you can't find that email, i can post it
again. Its solely about the fact that the document uses "IntServ" where it
should say (IntServ) "Guaranteed Service", because the second "IntServ" Service,
"Controlled Load" does not provide any determinisic latency guarantees (only GS does),
and therefore the unconstrained use of "IntServ" and not mentioning of "Guaranteed Services"
will be confusing to anyone who digs deeper into the IETF terminology (the refernce to rfc2212 is correct though).


That’s right! In the new version, we use the term “Guaranteed-Service IntServ”.

I think also tha he formulas in rfc2211 are more complex than what you write in 6.5,
i do like he simplification of 6.5, but it might be worth noting explicitly
hat those are implifications. specifically there are a bunch more parameters in
TSPEC/RSPEC than whats written in 6.5, and those parameters have impact on the
bounded latency in GS.

In Section 6.5, I added a sentence to make it more explicit.



Best,
Ehsan



--
Ehsan Mohammadpour
PhD Candidate at Swiss Federal Institute of Technology (EPFL)
IC IINFCOM, LCA2, INF 011, Station 14, 1015 Lausanne, Switzerland
https://people.epfl.ch/ehsan.mohammadpour<https://people.epfl.ch/ehsan.mohammadpour?lang=en>




On 8 Apr 2021, at 01:19, Toerless Eckert <[e-mail]<mailto:[e-mail]>> wrote:

Ehsan, *:

Thanks a lot for Section 7 in the latest draft version. Helps a lot to get
the linkage between this doc and the DetNet architecure an QoS options.

Alas, i do not see the concerns i raised with the authors about section 6.5 solved
that i send before IEF 110 in email. If you can't find that email, i can post it
again. Its solely about the fact that the document uses "IntServ" where it
should say (IntServ) "Guaranteed Service", because the second "IntServ" Service,
"Controlled Load" does not provide any determinisic latency guarantees (only GS does),
and therefore the unconstrained use of "IntServ" and not mentioning of "Guaranteed Services"
will be confusing to anyone who digs deeper into the IETF terminology (the refernce to rfc2212 is correct though).

I think also tha he formulas in rfc2211 are more complex than what you write in 6.5,
i do like he simplification of 6.5, but it might be worth noting explicitly
hat those are implifications. specifically there are a bunch more parameters in
TSPEC/RSPEC than whats written in 6.5, and those parameters have impact on the
bounded latency in GS.

Cheers
   Toerless

On Thu, Mar 25, 2021 at 07:56:01AM +0000, Mohammadpour Ehsan wrote:
Hello Lou,

Thanks for the comment, you???re right! I will use the following sentence:

"Then they are aggregated into eight macro flows based on their DetNet flow aggregate."

Best,
Ehsan


P.S. The intention of "Then, the input flows are identified using the information in (Section 5.1 of [RFC8939])???, was to say that we use Section 5.1 of RFC8939 to identify the individual DetNet flows.



--
Ehsan Mohammadpour
PhD Candidate at Swiss Federal Institute of Technology (EPFL)
IC IINFCOM, LCA2, INF 011, Station 14, 1015 Lausanne, Switzerland
https://people.epfl.ch/ehsan.mohammadpour<https://people.epfl.ch/ehsan.mohammadpour?lang=en><https://people.epfl.ch/ehsan.mohammadpour%3Chttps://people.epfl.ch/ehsan.mohammadpour?lang=en%3E>

On 22 Mar 2021, at 16:56, Lou Berger <[e-mail]<mailto:[e-mail]><mailto:[e-mail]>> wrote:


Ehsan

Thank you for the update.  I did a quick skim and it looks much improved.  I found one bit of new language confusing:

  Section 6.4

  In the considered queuing model, we considered the four traffic
  classes (Definition 3.268 of [IEEE8021Q]): control-data traffic
  (CDT), class A, class B, and best effort (BE) in decreasing order of
  priority.  Flows of classes A and B are together referred as AVB
  flows.  This model is a subset of Time-Sensitive Networking as
  described next.

  ....

  Then,
  the input flows are identified using the information in (Section 5.1
  of [RFC8939]).  Then they are aggregated into eight macro flows based
  on their traffic classes.  We refer to each macro flow as a class.

So in the first paragraph, you say "traffic classes" is defined by 802.1Q.  in the second, you say "their traffic classes" is defined based on [RFC8939].  I believe these are inconsistent definitions.  I think you are trying to say that

   Then they are aggregated into eight macro flows based on their *DetNet flow aggregate.*

This is assuming that the node isn't doing some sort of independent mapping of micro-flows to macro-flows.

Please let me know if I'm missing something.

Lou

On 3/22/2021 11:33 AM, Mohammadpour Ehsan wrote:

Dear WG,

We submitted a new version of the Bounded Latency draft that addresses the comments received in the list (https://datatracker.ietf.org/doc/html/draft-ietf-detnet-bounded-latency-04<https://tools.ietf.org/html/draft-ietf-detnet-bounded-latency-03><https://datatracker.ietf.org/doc/html/draft-ietf-detnet-bounded-latency-04%3Chttps://tools.ietf.org/html/draft-ietf-detnet-bounded-latency-03%3E>); it is now ready for submission to the IESG for publication.

Thanks Lou for his comments. In the new version, we address the comments as follows:

- Abstract
I think the abstract should to be updated to match the current content of the document. I suggest just copying the 4th and 5th paragraphs from the Introduction section.


Thanks for the suggestion. The new version is updated accordingly.

- general comment: classes of (DetNet) Service

The document states that it uses terminology from RFC8655, it also uses the terms "Classes of DetNet Service", " DetNet Classes of Service"  and other similar forms of these terms.  RFC8655 uses this term in one place " Class-of-Service schemes (e.g., DiffServ)" and also mentions "classes of DetNet flows"  in two places.  The data plane documents [RFC8964] and [RFC8934] also describe Class of Service in the context of DiffServ (and not DetNet).

In reading the document it seems that it is not using CoS in the typical way used in the IETF  or the other DetNet RFCs, and is introducing a new definition for CoS.   II think having a document specific definition of CoS probably isn't helpful. It would be best to use existing DetNet terminology.  I suspect  "flow aggregate" is the closest, but perhaps I'm missing something.


Thanks for your comment. In the new version, we used the term ???aggregate??? instead of class in section 3.1 and classes of DetNet flows. Section 6.4 uses the term ???traffic class??? as in IEEE802.1Q.

- alignment with the Data Plane RFCs

The lists in section 3.1 and in section 6.4 are not well aligned with the Data plane documents.  These sections should be aligned with the provisioning of IP flows, as summarized in Section  6 of RFC8939, and MPLS as summarized in Section 5 of RFC8964.

Thanks for your comment. Section 3.1 is about flow admission paradigm and not "flow creation???. The title is changed to ???flow admission??? to avoid possible confusion. In fact, this section is aligned with the DataPlane RFCs and in parallel with provisioning of IP flows. We modified section 6.4 to follow the Data plane documents.


- minor comment service vs frame preemption

Section 3.1 uses the term "un-provisioning", I read the text as meaning "service preemption".

In the rest of the document preemption refers to "frame preemption" and this should be clarified.

Thanks for the suggestion. In the new version, we use the term ???service preemption??? and update the term ???preemption??? to ???frame preemption???.


- id nits, please ensure the document passes idnits without errors, see

https://www6.ietf.org/tools/idnits?url=https://www.ietf.org/archive/id/draft-ietf-detnet-bounded-latency-02.txt

Thanks for your comment about idnits. We added the missing sections, i.e., ???security considerations??? and ???IANA considerations??? to the draft. Also resolved other comments by idnits.

Best,
Ehsan




--
Ehsan Mohammadpour
PhD Candidate at Swiss Federal Institute of Technology (EPFL)
EPFL EDIC PhD student representative
IC IINFCOM, LCA2, INF 011, Station 14, 1015 Lausanne, Switzerland
https://people.epfl.ch/ehsan.mohammadpour<https://people.epfl.ch/ehsan.mohammadpour?lang=en><https://people.epfl.ch/ehsan.mohammadpour%3Chttps://people.epfl.ch/ehsan.mohammadpour?lang=en%3E>


On 21 Feb 2021, at 23:15, Lou Berger <[e-mail]<mailto:[e-mail]><mailto:[e-mail]>> wrote:

All,

The WG last call is complete.

Authors,

Please bring any resolutions to comments received to the list. Once all comments are addressed please update the draft and let the WG know that it is ready for submission to the IESG for publications.

Thank you!

Lou

On 2/5/2021 8:30 AM, Lou Berger wrote:
All,

This starts working group last call on draft-ietf-detnet-bounded-latency-02
https://datatracker.ietf.org/doc/draft-ietf-detnet-bounded-latency/

The working group last call ends on February 19th.
Please send your comments to the working group mailing list.

Please note, there was an IPR disclosure against this document
submitted on January 14, 2021, see https://datatracker.ietf.org/ipr/4581/
While this disclosure came very late in the document life cycle,
it appears the filing was also relatively recent (2019-10-16) and
after the pre adoption IP call:
https://mailarchive.ietf.org/arch/msg/detnet/1mwYdadDzTLudcuH5T5O7R_KlDs

Positive comments, e.g., "I've reviewed this document
and believe it is ready for publication", are welcome!
This is useful and important, even from authors.

Thank you,
Lou (DetNet Co-Chair & doc Shepherd)



_______________________________________________
detnet mailing list
[e-mail]<mailto:[e-mail]><mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/detnet



_______________________________________________
detnet mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/detnet


--
---
[e-mail]<mailto:[e-mail]>