Hi Jie,

Thanks for your reply. Please see inline comments at [Fan].

Regards,
Fan

On Wed, Dec 3, 2025 at 11:09 AM Dongjie (Jimmy) <[email protected]> wrote:

> Hi Fan,
>
>
>
> Thanks for your detailed review of this draft. They are helpful for
> polishing the document.
>
>
>
> Please see some replies inline:
>
>
>
> *From:* Fan Zhang <[email protected]>
> *Sent:* Tuesday, December 2, 2025 11:24 AM
> *To:* [email protected]
> *Cc:* RTGWG <[email protected]>; FANTEL <[email protected]>
> *Subject:* [fantel] Some comments on
> draft-dong-fantel-problem-statement-02
>
>
>
> Hi authors,
>
> Thanks for the draft. It's good to have a draft that helps clarify the
> problems and scope of FANTEL. I've read it and have some comments below.
>
> [Comments]
>
> 1) "fast network notification" and "fast notification" are both used in
> the document. Is there any difference between them? If not, it may be
> better to align the terminology.
>
> [Jie] Thanks for pointing this out. I agree it is better to align the
> terminology. IMO this is a good question and we would like to hear opinions
> from participants on both the RTGWG and FANTEL lists:
>
> [Jie] "fast notification" is what was firstly used in FANTEL BoF, "network
> notification" is what we used in initial versions of this draft following
> the guidance from our AD, and  "fast network notification" is what we hear
> people proposing during the RTGWG meeting at IETF 124.
>
> [Jie] Which one should we use in future version of this draft and other
> related drafts? We are open to feedbacks.
>
[Fan] Just my personal view. "fast network notification" might be clearer
in indicating that the notifications are propagated across the network.

>
> 2) Section 3-Overhead and Scalability Challenges: Telemetry usually works
> in a publish/subscribe mode, rather than "flooding". "telemetry streaming"
> may be better.
>
> [Jie] I agree telemetry streaming is more accurate, thanks for the
> suggestion, will fix it in next revision.
>
>
> 3) Section 4: "before congestion or micro-loops develop". "develop" might
> imply a gradual progress. Perhaps "occur" would be clearer.
>
> [Jie] My personal opinion is develop may be better here. “develop”
> indicates the congestion and micro-loop can become more significant if it
> takes longer time to send the notification. With fast notification and
> action, it is possible to avoid severe congestion and micro-loop, while
> they may still exist for a short period.
>
[Fan] Got it. I understand your intention to imply more severe consequences
if congestion or micro-loops persist for a longer time. So you might want
to avoid a possible misunderstanding like "before congestion or micro-loops
occur", which could sound too strict about the notification timing. Perhaps
"further develop" would be clearer?

>
> 4) Section 4.1.1-BFD: "nor can it notify the failure or congestion to
> other nodes". IMHO, BFD can trigger the local IGP process to notify the
> failure to other nodes.
>
> [Jie] Yes BFD can be used for the detection of IGP neighbor failure, then
> IGP will act accordingly and propagate state updates to other nodes in the
> network. While BFD itself is not used as a notification to other nodes
> about the failure. Also BFD is not used for the notification of congestion
> or other state changes.
>
> 5) Section 4.1.2: "Fast notification alerts upstream nodes". I agree that
> it's obvious that upstream nodes might need to be notified in some cases.
> But I'm wondering whether there are other cases where fast notification
> might alert other nodes (such as neighbors)?
>
> [Jie] In the example described in section 4, the notification is used to
> notify the nodes which are sending traffic towards the failure point, so as
> to shift the traffic to alternate link/path, thus from the perspective of
> traffic flow, the notification needs to be sent to the upstream nodes.
> Depending on the traffic flows to be shifted, the notification may be sent
> to different nodes (which are considered as the upstream nodes of the
> traffic flow under consideration). We will consider how to make this
> clearer.
>
>
>
> [Nits, take it or leave it]
>
>
>
> 1) Section 4.1: "exchange terabits per second of data" --> "exchange
> terabits of data per second"
>
> 2) Section 4.1.1-FRR: "The former lacks of global network situation" -->
> "The former lacks a global view of the network state"
>
> 3) Section 4.1.1: "may already be missed" --> "may have already been
> missed"
>
> 4) Section 5.2: "protocols, analytics, NMS, etc." --> "protocols,
> analytics systems, NMS, etc."
>
> 5) Section 5.4: "but not limit to" --> "but not limited to"
>
> 6) Section 5.4: "Switch all traffic" --> "Switches all traffic". Might
> need to align with other bullets.
>
> 7) Section 6: "too slow, too coarse, or too resource-intensive" --> "too
> slow, too coarse, or too resource-intensive to ......" Might need to
> clarify the specific consequences.
>
>
>
> [Jie] Thanks for catching the nits, they will definitely help in improving
> the readability of this document.
>
>
>
> Best regards,
>
> Jie
>
>
>
> Regards,
>
> Fan
>
_______________________________________________
rtgwg mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to