Re: [Detnet] Some more Comments on detnet-yang-15 Re: [Last-Call] Yangdoctors last call review of draft-ietf-detnet-yang-14

[Don Fedyk <[e-mail]>]

Hi Tom

Thanks for the comments. See Inline [Don]
I have pushed updates into https://github.com/detnet-wg/draft-ietf-detnet-yang

Cheers
Don

-----Original Message-----
From: tom petch <[e-mail]> 
Sent: Tuesday, December 28, 2021 6:09 AM
Cc: [e-mail]; [e-mail]
Subject: Some more Comments on detnet-yang-15 Re: [Last-Call] Yangdoctors last call review of draft-ietf-detnet-yang-14

On 23/12/2021 11:40, tom petch wrote:
Also

"IP six tuple"
perhaps 'sextuple' That said, I do not know what the six are and cannot find a reference to six in either RFC8939 or RFC8964

[Don]It is 6-tuple in RFC8939.  We can use the exact same spelling to make it easier to find.   


"   Outgoing is viewed as going up the stack from"
Looking at fig 1 I would have assumed the opposite, going down the 
stack; perhaps IETF versus IEEE.
[Don] The diagram Figure 1 is consistent with the architecture. 
Down is pointing from the application. So, OK. Yes, I the text "up the stack should" 
be "down the stack" to be consistent with the diagram and with the architecture.   



" enum copy-from-app-flow {
     description
       "This type means copy the app-flow sequence number to the "
Mmm it looks like an enum to me and not a type
[Don] Agree. 

"  enum generate-by-detnet-flow {
      description
         "This type means generate the sequence number by the "
Ditto
[Don] Agree.

"     grouping detnet-flow-spec {
        choice detnet-flow-type {
          case ip-detnet-flow {
          case mpls-detnet-flow { "
elsewhere the case are 'ip' and 'mpls'; is this different?
[Don] Yes this is a bit different. For IP, this refers to IP as a DetNet flow identifier.
It uses an IP - wild card capable grouping. This allows Aggregation with IP headers.
 
In the other cases it is IP forwarding. For a specific IP header.
It uses a specific IP address grouping. 

For MPLS it is the same - Matching an MPLS DetNet flow (a label/label stack)
You could be matching just an outer label. 
Compared to the other cases using an MPLS label/label stack. MPLS is more subtle than the IP case where prefixes are used but it parallels the IP cases.

I think the distinction was deliberate and is useful.
I will leave this as is.   
 

      "RFC 9016 Section 5.5, IEEE802.1Q";
I would find this clearer with the two references on two separate lines
[Don] Sure. 


"       F-label for DetNet service or service-to-forwarding "
Perhaps a reference to "RFC 8964 Section 4.2.3";
[Don] Sure.

Tom Petch


> Borrowing a useful cc: with a related topic
>
> two documents in the YANG module need adding to the I-D References
> RFC8938
> IEEE 802.1Qcx-2020
>
> contact URL
> insecure, obsolete - https://datatracker ...
>
> Editor: Mach Chen
> not in the authors of the I-D
>
> TLP out of date
>
> Reference: this ID when published
> Reference: "RFC XXXX: Deterministic Networking (DetNet) YANG Model
>
>
> Tom Petch
>
>
> On 09/11/2021 20:06, Xufeng Liu via Datatracker wrote:
>> Reviewer: Xufeng Liu
>> Review result: Ready with Nits
>>
>> Thanks to authors for addressing the previous review comments.
>>
>> The updates look good. The followings are a few additional nits:
>>
>> 1) In the model, “container flow-spec” has been changed to “container
>> traffic-spec”, but the description has not been updated, shown as below:
>>
>>           container traffic-spec {
>>             description
>>               "Flow-specification specifies how the Source transmits
>>                packets for the flow.  This is the promise/request of the
>>                Source to the network.  The network uses this flow
>>                specification to allocate resources and adjust queue
>>                parameters in network nodes.";
>>
>> 2) Most names of list and leaf-list have been fixes. The following
>> three were
>> missed: “leaf-list member-apps” should be “leaf-list member-app”
>> “leaf-list
>> member-services” should be “leaf-list member-service” “leaf-list
>> member-fwd-sublayers” should be “leaf-list member-fwd-sublayer”
>>
>> 3) Section 10.  Security Considerations would need to include a list of
>> “sensitive or vulnerable” nodes.  RFC 8349 shows an example.
>>
>> Thanks,
>> - Xufeng