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

["Scharf, Michael" <[e-mail]>]

Hi Camilo,

Sorry for the delay. Please see below.

> -----Original Message-----
> From: Camilo Cardona <[e-mail]>
> Sent: Thursday, July 28, 2022 7:21 PM
> To: Jeffrey Haas <[e-mail]>; Scharf, Michael <Michael.Scharf@hs-
> esslingen.de>
> Cc: [e-mail]; draft-ietf-tcpm-yang-
> [e-mail]; [e-mail]; [e-mail]
> Subject: Re: [GROW] How to reuse the tcp model in the BMP model - asking
> for suggestions
> 
> Hello all,
> 
> First of all, thanks to Jeff for giving us a bit more hints about how to use the
> tcp model. We’ll try to see how to leverage it as much as possible.
> 
> Michael,
> 
> The draft mentions Maximum Segment Size  in section 3.1, first point
> (“Typical examples are the Maximum Segment Size (MSS) or configuration
> related to hardware offloading.”) but I don’t see it anywhere in the model. I
> would assume that you don’t want to include it. Still, This is sort of confusing,
> at least to me.

The context of this sentence is:

  *  Interface configuration: It can be useful to use different TCP
      parameters on different interfaces, e.g., different device ports
      or IP interfaces.  In that case, TCP parameters can be part of the
      interface configuration.  Typical examples are the Maximum Segment
      Size (MSS) or configuration related to hardware offloading.

> Since I couldn’t fully understand what your goal with that example is, I cannot
> offer a suggestion for making it clear, however I think it would be nice if you
> take a look at that sentence.

As different interfaces can have a different MTU, it is not uncommen to use different MSS values for connections originating/terminating on different interfaces. At least the Linux kernel apparently picks the MSS per interface. As a result, it hardly makes sense to model in YANG a single MSS value for all interfaces. Instead, the MSS value would have to be per interface and set in a corresponding YANG model for the interface. Theoretically, we could do this for MSS, e.g., by augmentation. But for other parameters, such as hardware offload configuration, this would get complex and technology-specific. As a result, the proposed YANG model stays away from interface-specific parameters.

Some terminology in this section will be updated in the next revision to address the IESG feedback. As part of this, I could also add a sentence along the lines of

   "This document does not model interface-specific parameters"

to this section. Would that address your concern?

Note that I am actually not convinced that this specific section needs editing, but as we have to modify other text in that section anyway, I can try to tweak this wording as well.

Thanks

Michael



> Thanks,
> Camilo C
> 
> On 25/7/22, 14:12, "Jeffrey Haas" <[e-mail]> wrote:
> 
>     Camilo,
> 
>     Responding somewhat late to this thread, but I think that you will find that
>     the latest updates to the tcpm module addressed most of your concerns, if
>     not all of them.  The BGP YANG module, as Michael notes below, helped
> refine
>     some of the use case scenarios.
> 
>     The audit was primarily covering how to use authentication.  Some
> additional
>     focus on other TCP properties might be worth evaluating.
> 
> 
>     -- Jeff
> 
>     On Tue, Jun 21, 2022 at 09:12:36AM +0000, Scharf, Michael wrote:
>     > Hi Camilo,
>     >
>     > There are existing examples for YANG modules that model application-
> specific configuration for TCP connections, such as:
>     >
>     >   *   draft-ietf-idr-bgp-model
>     >   *   draft-ietf-netconf-tcp-client-server
>     >
>     > I would assume that BMP could be modeled like that.
>     >
>     > Note that the model for TCP-AO authentication has changed in draft-ietf-
> tcpm-yang-tcp-07 because of last call comments, i.e., some model aspects
> can still be subject to change.
>     >
>     > Michael
>     >
>     > From: Camilo Cardona <[e-mail]>
>     > Sent: Friday, June 17, 2022 4:37 PM
>     > To: Scharf, Michael <[e-mail]>; draft-ietf-tcpm-
> [e-mail]; [e-mail]
>     > Cc: [e-mail]; [e-mail]
>     > Subject: Re: How to reuse the tcp model in the BMP model - asking for
> suggestions
>     >
>     >
>     > Hello Michael,
>     >
>     >
>     >
>     > First of all, thanks for considering our questions and letting us know
> about this new version.
>     >
>     >
>     >
>     > Please keep in mind that the BMP model draft is in very early stages, not
> even a WG draft yet, it might change in the future. So, we apologise if we
> cannot give you exact requirements.
>     >
>     >
>     >
>     > Having said that, it will feasible that the model will need to include
> multiple TCP configurations. Defining the connection might be specific to the
> application, but BMP requires other  features like authentication, MSS,
> keepalives which seem general enough. What we wanted was to leverage
> other model for this, if existing.
>     >
>     >
>     >
>     > Thus, I guess the most general question we can make is , How is the
> recommended way of reusing the tcp model?
>     >
>     >
>     >
>     > Thanks,
>     >
>     > Camilo Cardona
>