[Lsr] 答复: 【Request AD Step In】 Working Group Adoption of "IGP Unreachable Prefix Announcement" - draft-ppsenak-lsr-igp-ureach-prefix-announce-04

[Aijun Wang]

Hi, John:

Thanks for your reply.
The key concerns for the issue is that although 
https://datatracker.ietf.org/doc/draft-wang-lsr-prefix-unreachable-annoucement 
has several advantages, it hasn't been given the adoption call until now. This 
is unfair and also our main appeal reason.

Considering there are two different approaches to solve some overlapping 
scenarios, I think we can consider to adopt both of them as WG documents, 
similar with actions of other WGs.
We can let the industry to select the final solution to implement and deploy 
within their networks.

We have enough energies to accomplish the final implementation and deployment.

Some detail responses are inline below.


Best Regards

Aijun Wang
China Telecom

-邮件原件-
发件人: forwardingalgori...@ietf.org [mailto:forwardingalgori...@ietf.org] 代表 John 
Scudder
发送时间: 2023年11月1日 6:02
收件人: Aijun Wang 
抄送: lsr ; draft-ppsenak-lsr-igp-ureach-prefix-annou...@ietf.org
主题: Re: [Lsr] 【Request AD Step In】 Working Group Adoption of "IGP Unreachable 
Prefix Announcement" - draft-ppsenak-lsr-igp-ureach-prefix-announce-04

Hi Aijun,

I apologize for the length of time it’s taken me to respond to your request. 

Having now taken the time to study the question properly, including a review of 
both drafts in question, the WG adoption call, and the subsequent email, here’s 
my take.

In large part, your position appears to be based on historical precedence — 
your draft was published first. (This is your “follower solution… initiator” in 
the email I’m responding to, as well as the first three “which draft is the 
first” points in your follow-up.) This is true of course. Furthermore, although 
our formal process does not take into account such questions as “who came 
first?” I think it would be safe for me to say that people generally do try to 
do not just what’s required, but what’s right, in terms of acknowledging prior 
work. For this reason, I was a little surprised to see no acknowledgment of the 
contributions of your draft in draft-ppsenak. But I think such an 
acknowledgment — which is a norm, not a requirement — is the most you can 
expect for having published the first draft that covers the same general 
subject area as draft-ppsenak. This might also be a good time to remind you 
that draft-wang-lsr-prefix-unreachable-annoucement-00 includes the statement,

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

I encourage you to review BCP 78 if you haven’t recently.
【WAJ】In contrast, we include the above statement from the version 00: 
https://datatracker.ietf.org/doc/html/draft-wang-lsr-prefix-unreachable-annoucement-00,
 and also acknowledge the comments from Peter Psenak, Les Ginsberg, Bruno 
Decraene, Acee Lindem, Shraddha Hegde, Robert Raszuk, Tony Li, Jeff Tantsura 
and Tony Przygienda for their suggestions and comments on draft-wang.

In short, I’m not persuaded by the first-to-publish argument.
【WAJ】We are not advocating the fist-to-publish argument, but the first should 
be considered first for adoption call, or else, there must be reasonable 
explanations for not doing so.

The other major point made by you, and others advocating for the consideration 
of draft-wang as the WG solution and against draft-ppsenak, is that draft-wang 
is said to cover more cases. (This is “cover more scenarios” in your email, as 
well as point five, “cover more scenarios” in your follow-up.) There was some 
spirited debate about whether the draft does so successfully, or not, but I 
don’t want to take a position on that in this email. Rather, what I observe is 
that since these points were made clearly, and repeatedly, in the WG adoption 
email thread as well as at other times previously, it can’t be argued that the 
WG didn’t know that draft-wang claims to address (for example) area partition, 
and that draft-ppsenak explicitly doesn’t. So, this suggests those who 
supported the adoption of draft-ppsenak either implicitly, or explicitly, 
believed that the additional use cases draft-wang claims to address are not 
important. At least, not important to address in this draft, at this time, as 
part of this adopted WG work.
【WAJ】Prefix unreachable announcement is one general mechanism, the solution 
shouldn't be limited only on some narrow scopes. For standardization work, we 
should look further.

In your follow-up, you also proposed that “which explicit signaling mechanism 
is simpler” should be a criterion (point four). In my experience, this kind of 
question seldom leads to a useful outcome since it’s so subjective. I will say 
however that many of the people who responded to the WG adoption call made it 
clear they had such considerations in mind, so I think there is good reason to 
think the WG has taken this question into account.
【WAJ】The adoption call is issued only for one approach, not both of them, then 
how can we get the above conclusions?

I also want to speak to the questions of whether the WG 

[Lsr] Agenda Re: IETF 118 AIDC Side Meeting

[Yingzhen Qu]

Hi all,

The agenda for the AIDC side meeting is now available at: AIDC-IETF118/AIDC
Side Meeting Agenda.md at main · Yingzhen-ietf/AIDC-IETF118 (github.com)


We have presentations from both industry and academia, covering topics from
fundamental requirements to congestion control in Data Center networks for
AI.

For remote attendees, please use the webex link:
https://ietf.webex.com/meet/ietfsidemeeting2

Meeting Time: Tuesday, 7 Nov, 17:00 - 19:00 *(Central European Time -
Prague)*
Location: Palmovka 1/2
Meeting Materials: Yingzhen-ietf/AIDC-IETF118: Meeting materials for the
AIDC side meeting at IETF 118. (github.com)


If you have any questions, please let us know.

Thanks,
Jeff and Yingzhen

On Sun, Oct 22, 2023 at 10:37 PM Yingzhen Qu 
wrote:

> Hi,
>
> (RTGWG Chair hats off)
>
> Following our productive discussions at IETF 117's AIDC side meeting, we
> are excited to announce another meeting during IETF 118 in Prague. This
> gathering will serve as a continuation of the ongoing dialogue.
>
> The primary focus of this meeting is to delve into the introduction of new
> technologies within large-scale data centers, especially in the context of
> AI model training and related applications. While we'll give special
> attention to this theme, we welcome discussions on a broader spectrum of
> topics.
> Our objectives for this meeting include:
> • Gaining insights into real-world use cases
> • Identifying emerging networking challenges and requirements
> • Exploring innovative research and standardization opportunities
>
> We look forward to engaging in these discussions and collectively
> advancing our understanding of the evolving landscape of data center
> networking for AI. Your active participation will be highly valuable as we
> work towards addressing these challenges and fostering innovation in this
> domain.
>
> The detailed agenda, slides and link for remote attendees will be uploaded
> to the GitHub repository soon.
>
> Meeting Time: Tuesday, 7 Nov, 17:00 - 19:00
> Location: Palmovka 1/2
> Meeting Materials: Yingzhen-ietf/AIDC-IETF118: Meeting materials for the
> AIDC side meeting at IETF 118. (github.com)
> 
>
> If you have any questions or comments, please contact us.
>
> Thanks,
> Jeff and Yingzhen
>
___
Lsr mailing list
Lsr@ietf.org
https://www.ietf.org/mailman/listinfo/lsr

Re: [Lsr] 【Request AD Step In】 Working Group Adoption of "IGP Unreachable Prefix Announcement" - draft-ppsenak-lsr-igp-ureach-prefix-announce-04

[John Scudder]

Hi Aijun,

I apologize for the length of time it’s taken me to respond to your request. 

Having now taken the time to study the question properly, including a review of 
both drafts in question, the WG adoption call, and the subsequent email, here’s 
my take.

In large part, your position appears to be based on historical precedence — 
your draft was published first. (This is your “follower solution… initiator” in 
the email I’m responding to, as well as the first three “which draft is the 
first” points in your follow-up.) This is true of course. Furthermore, although 
our formal process does not take into account such questions as “who came 
first?” I think it would be safe for me to say that people generally do try to 
do not just what’s required, but what’s right, in terms of acknowledging prior 
work. For this reason, I was a little surprised to see no acknowledgment of the 
contributions of your draft in draft-ppsenak. But I think such an 
acknowledgment — which is a norm, not a requirement — is the most you can 
expect for having published the first draft that covers the same general 
subject area as draft-ppsenak. This might also be a good time to remind you 
that draft-wang-lsr-prefix-unreachable-annoucement-00 includes the statement,

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

I encourage you to review BCP 78 if you haven’t recently.

In short, I’m not persuaded by the first-to-publish argument.

The other major point made by you, and others advocating for the consideration 
of draft-wang as the WG solution and against draft-ppsenak, is that draft-wang 
is said to cover more cases. (This is “cover more scenarios” in your email, as 
well as point five, “cover more scenarios” in your follow-up.) There was some 
spirited debate about whether the draft does so successfully, or not, but I 
don’t want to take a position on that in this email. Rather, what I observe is 
that since these points were made clearly, and repeatedly, in the WG adoption 
email thread as well as at other times previously, it can’t be argued that the 
WG didn’t know that draft-wang claims to address (for example) area partition, 
and that draft-ppsenak explicitly doesn’t. So, this suggests those who 
supported the adoption of draft-ppsenak either implicitly, or explicitly, 
believed that the additional use cases draft-wang claims to address are not 
important. At least, not important to address in this draft, at this time, as 
part of this adopted WG work.

In your follow-up, you also proposed that “which explicit signaling mechanism 
is simpler” should be a criterion (point four). In my experience, this kind of 
question seldom leads to a useful outcome since it’s so subjective. I will say 
however that many of the people who responded to the WG adoption call made it 
clear they had such considerations in mind, so I think there is good reason to 
think the WG has taken this question into account.

I also want to speak to the questions of whether the WG adoption decision was 
too hasty, whether there should be more deliberation in the WG, and whether 
there should have been a separate adoption call for draft-wang, which are 
points you’ve made emails other than the one I’m replying to. Regarding whether 
it was too hasty — as you say in this email, this work has been in progress 
since 2019. The merits of the solutions have been debated extensively. A 
considerable amount of valuable WG meeting time has been devoted to these 
discussions, as well as a great many emails. It’s hard for me to see the WG 
adoption decision as being made without due deliberation — the opposite if 
anything. Regarding whether there should have been an adoption call for 
draft-wang — our process allows considerable latitude to WG chairs in how they 
choose to run these things. In reviewing this adoption call, it seems to me 
that all participants were clear that in practice and regardless of what the 
subject line was, they were really addressing a multi-part question: should the 
WG work on this area? If so, should the base document be draft-ppsenak, or 
draft-wang? These questions received a full airing, as far as I can tell.

As you know, the IETF runs on “rough consensus”. This is true for WG adoptions 
just as for anything else, and it sometimes requires WG chairs to make hard 
decisions to call a consensus where some WG contributors are “in the rough”. 
After reviewing the WG adoption call, drafts, and history, it appears to me 
that the WG chairs have listened to all the positions put forward and 
considered them, and judged the rough consensus to favor the adoption of 
draft-ppsenak. I don’t see sufficient evidence to make me believe I should 
overrule the WG chairs’ judgment.

Finally, I will point out that you have many options still open to you if you 
strongly feel that the scenarios that are not covered by the adopted document 
are crucial. 

Thanks for your patience as I