Re: [tcpm] How to reuse the tcp model in the BMP model - asking for suggestions

[t petch <[e-mail]>]

On 04/05/2022 18:22, Scharf, Michael wrote:
> Hi Camilo,
>
> Thanks a lot for reaching out!
>
> We have received quite a number of other comments from the directorate reviews. As a result, some parts of the model will probably have to be modified. We will have a careful look at your suggestion as well.
>
> Regarding the 4-tuple connection, at first sight this is seems specific to your use case. If so, your own container is probably the simplest solution. I don’t think this concept generically applies to all uses of TCP.
>
> The authors will have a look at the definition of groupings to simplify re-use by your model.

and to complicate it for most other users.  A YANG grouping, like a 
procedure or subroutine, can make sense when it is well-defined, its 
meaning or function self-evident.  Most grouping in IETF models do not 
qualify and mean the user has to jump backwards and forwards, perhaps in 
and out of multiple documents, while losing the thread of what it is 
that is being modelled.

YANG has augment, a much superior way of building on the work of others 
so adding additional node as augment points is useful, carving a model 
up into disjoint pieces less so.

Tom Petch





We will get back to you on that.
>
>
>
> Thanks
>
>
>
> Michael
>
>
>
>
>
>
>
> From: Camilo Cardona <[e-mail]>
> Sent: Tuesday, May 3, 2022 5:40 PM
> To: [e-mail]; [e-mail]
> Cc: [e-mail]; [e-mail]
> Subject: How to reuse the tcp model in the BMP model - asking for suggestions
>
>
>
> Hello draft-ietf-tcpm-yang-tcp authors,
>
>
>
> We are analysing how to leverage the tcpm model for managing tcp connectivity in the BMP yang model (draft-cptb-grow-bmp-yang-02).
>
>
>
> In general, how do you suggest other models reuse ietf-tcpm? Most of tcpm model is defined directly in containers, not as groups, and we are finding hard to take the pieces that we need.
>
>
>
> Let us shortly gointo details.
>
>
>
> Regarding the 4-tuple connection, in BMP, we would like to have a “default” case in which the configured device connects to another using ip/port (i.e. active case),  but also to provide the option of expecting the  tcp connection over a local ip/port (i.e. passive case). We might need our own container for this.
>
>
>
> However, we would like to integrate the “common”, and “statistics” sections of ietf-tcpm, but, they are containers, so we really do not know what to do besides recreating it.
>
>
>
> Do you have any suggestions?
>
>
>
> Thanks,
>
> draft-cptb-grow-bmp-yang authors
>
>
>
>
>
>
> _______________________________________________
> tcpm mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/tcpm
>