Re: [Teas-ns-dt] Figure in the ACTN Applicability section

[Eric Gray <[e-mail]>]

John and Dhruv,



              Actually 0 or more "SBIs" _MAY_ exist.



              We should not discuss how the TSC produces its magic, except loosely and in logical terms.



              It is fairly easy to envision a transport slice service that is provided by a single physical box, or even a single virtual function within a box that provides multiple such functions, possibly among others.



              In the event that this is the case, and we certainly have no reason to exclude this possibility, there would quite likely be no SBI at all (what exactly would be the point?).



              Conversely, a TSC implementation might be conceived (and implemented) that uses entirely proprietary mechanisms to provision the underlying network infrastructure.



              Again, we cannot rule this possibility out, and again there would be no SBI (or at least none specified by the IETF).



              That is not to say that either of these implementation examples would not offer visibility to appropriate model information.  Such could be done in any case where the appropriate standard model can be mapped to what the TSC actually does.  This is why a logical model may be defined, but does not guarantee that services are exactly realized as described by that model.



              As a result, we do not – at least in the framework – need to delve very deeply into how the TSC _MAY_ use any form of SBI as, doing so at this point would likely restrict implementation options unnecessarily.



              Also see embedded replies below…



--

Eric



-----Original Message-----
From: John E Drake <jdrake=[e-mail]>
Sent: Thursday, May 14, 2020 1:46 PM
To: Dhruv Dhody <[e-mail]>; Eric Gray <[e-mail]>
Cc: haomianzheng <[e-mail]>; Luis M. Contreras <[e-mail]>; Xufeng Liu <[e-mail]>; Kiran Makhijani <[e-mail]>; Belotti, Sergio (Nokia - IT/Vimercate) <[e-mail]>; [e-mail]
Subject: RE: [Teas-ns-dt] Figure in the ACTN Applicability section
Importance: High



Dhruv,



From my perspective, there are going to be a multiplicity of SBIs, some of which may already exist.



Yours Irrespectively,



John





--- [SNIP - Juniper disclaimer removed] ---



> -----Original Message-----

> From: Teas-ns-dt <[e-mail]<mailto:[e-mail]>> On Behalf Of Dhruv

> Dhody

> Sent: Thursday, May 14, 2020 1:38 PM

> To: Eric Gray <eric.gray=[e-mail]<mailto:eric.gray=[e-mail]>>

> Cc: haomianzheng <[e-mail]<mailto:[e-mail]>>; Luis M. Contreras

> <[e-mail]<mailto:[e-mail]>>; Xufeng Liu <[e-mail]<mailto:[e-mail]>>;

> Kiran Makhijani <[e-mail]<mailto:[e-mail]>>; Belotti, Sergio (Nokia -

> IT/Vimercate) <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>

> Subject: Re: [Teas-ns-dt] Figure in the ACTN Applicability section

>

> --- [SNIP - Ericsson disclaimer removed] ---

>

>

> Hi Eric, DT,

>

> Lets continue the discussion from the call -

>

> - I agree with the fact that everything above TSC (and TSC-NBI) can be

> a big box and we don't have to worry about it in this I-D.

> - Everything below TSC (and TSC-SBI) needs a little more details as

> this is something that belongs in the IETF.



It is not in our charter to “provide details” for everything that “belongs to the IETF.”



We can – and should – provide examples, where possible, but not as an exhaustive list.



Unless doing so is necessary to highlight specific issues that we need to resolve, in this project.



Are you arguing that is the case?



> The TSC-SBI is a TS

> realization interface and could have different granularity based on

> the network controller it is interfacing with.



This will only be the case if we  are unable to define the NBI abstraction, and the TSC, in such a way that they hide the way that the TSC uses any presumed SBI – including the possibility that the TSC does not explicitly use any SBI.



Also, how the TSC realizes any transport slice is explicitly out of scope in this project.



> But we need to clarify  that it is NOT the customer service model (L3SM etc).



We only need to clarify this to the extent that we might describe any service model in any detail.



A network operator _MAY_ certainly use L3SM models to manage their L3 services.  But what we are working on is a higher layer abstraction that – itself – _MAY_ uses L3SM models, L2SM models, other models, or no models at all.



> - We can further discuss if describing things in terms of RFC 8309

> models is useful, I saw some usefulness at least to drive the discussion.

> - I agree with Eric, that a figure is useful. Perhaps we could think

> about removing the arrows towards ACTN boxes, if that helps!



That is essentially the figure that I suggested (look at it again).



I am hard-pressed to argue objectively about including a figure.  I do not personally like them, because (as Reza intimated) pictures can (and often do) leave a lo open to (possibly incorrect) interpretation.



If anything, I tend to include too few figures.



But I do not think that we are at fault if people base their conclusions entirely on a figure, without carefully reading text that is included to clarify the meaning of the figure.  At least not unless the figure is highly misleading, or the text is not consistent with the figure – as either of these situations will inevitably produce more confusion than clarity.



And, as I said during the meeting, a lot of folks simply _MUST_ have a figure.  The adage they cite is “A picture is worth a thousand words.”  As a rule, as implied above, I am exactly the opposite.  I reply “Give me the thousand words, or you wasted time creating the picture.”



I suspect that others are like me in that the only thing I am naturally inclined to conclude from a picture is whether or not it is esthetically pleasing.



In my experience, if you give a “pictures please” person a purely textual explanation, they will immediately start drawing a figure – which they will then need to confirm with the authors is correct – hence drawing out the intended communication process.  You can easily determine that you are dealing with such a person, because A) they want to draw a figure, and B) they likely will not get it right the first one or two times they try.



It takes all kinds.



If the folks in the design team feel that the mapping of this work to the work done in ACTN is not complicated enough to warrant a figure, I am willing to take it out.



>

> Thanks!

> Dhruv

>

> On Thu, May 14, 2020 at 8:28 PM Eric Gray

> <eric.gray=[e-mail]<mailto:eric.gray=[e-mail]>> wrote:

> >

> > Folks,

> >

> >

> >

> > Obviously, we need to re-assess what to include in the figure.

> >

> >

> >

> > In an intermediate figure, I had tried having the MDSC as “floating”

> > between

> the TSC and the PNC, because (as is clear in these discussions)

> exactly what the mapping between TSC and specific ACTN entities is the

> main ambiguity issue in the current section.

> >

> >

> >

> > But there is a bigger issue that is also clear in these discussions:

> > some of the

> areas for minor disagreement in the pictures is in those parts of the

> figure that are in the “we don’t exactly care” region.

> >

> >

> >

> > It is not material (in the Framework draft, at least – but probably

> > in all of the

> work in scope for the DT) exactly how we define “customer,” (hierarchy

> of) “orchestration,” and other entities above the point where <_some_

> _entity_> issues transport slices requests via the TSC NBI.

> >

> >

> >

> > Any of these things may be the “user” of the TSC NBI, and prolonged

> > and

> detailed discussion about how that happens, or who is who in any

> process for which the access to the NBI is used, is pretty much

> irrelevant and a waste of time.

> >

> >

> >

> > Similarly, discussion of what goes on “under the hood” of the TSC

> > itself is

> likewise not tremendously interesting.

> >

> >

> >

> > We had agreed that the TSC portion of the figure is useful to the

> > extent that it

> clarifies logical relationships between the TSC and the logical

> entities that MAY exist between the TSC and the underlay network it

> uses to realize transport slices, as well as the logical entities that

> MAY exist between a customer (for some definition of “customer”), an

> arbitrary “hierarchy” of 0 or more layers of “orchestration,” and the TSC NBI.

> >

> >

> >

> > We also need to talk about at least an outline of a mapping to

> > existing work

> on ACTN and the work we are doing in the NS-DT.

> >

> >

> >

> > I am personally comfortable with doing that entirely without a

> > figure, but

> experience tells me that many folks are not able to get as much out of

> any explanation that does not include a figure or two.

> >

> >

> >

> > At this point, I would prefer to revisit the figure with the

> > floating relationship

> between TSC and ACTN – and follow that up with text that describes

> essentially what I have included above.

> >

> >

> >

> > That figure was as follows:

> >

> >

> >

> >        +------------------------------------+

> >

> >        |             Customer               |  |

> >

> >        +------------------------------------+        ACTN

> >

> >                          A                     |  Terminology

> >

> >                          |                        and Concepts

> >

> >                          V                     |

> >

> >        +------------------------------------+

> >

> >        |      A highter level system        |  |   +-----+

> >

> >        |(e.g e2e network slice orchestrator)| ===> | CNC |

> >

> >        +------------------------------------+  |   +-----+

> >

> >                          A                            A

> >

> >                          | TSC NBI             |      |

> >

> >                          V                            |

> >

> >        +------------------------------------+  |      | CMI

> >

> >       |      Transport Slice Controller    |         V

> >

> >        +------------------------------------+  |   +------+

> >

> >                          A                         | MDSC |

> >

> >                          | TSC SBI             |   +------+

> >

> >                          | TSC SBI                    A

> >

> >                          |                     |      | MPI

> >

> >                          V                            V

> >

> >        +------------------------------------+  |   +-----+

> >

> >        |        Network Controller(s)       | ===> | PNC |

> >

> >        +------------------------------------+  |   +-----+

> >

> >

> >

> >                 Terminology/Concepts           |

> >

> >                Used in this Document

> >

> >                                                |

> >

> >

> >

> > --

> >

> > Eric

> >

> >

> >

> > From: Xufeng Liu <[e-mail]<mailto:[e-mail]>>

> > Sent: Thursday, May 14, 2020 9:29 AM

> > To: Luis M. Contreras <[e-mail]<mailto:[e-mail]>>

> > Cc: Eric Gray <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>

> > Subject: Re: [Teas-ns-dt] Figure in the ACTN Applicability section

> >

> >

> >

> > Luis's diagram makes more sense to me. Some additional thoughts:

> >

> >

> >

> > 1) Transport Slice Controller (TSC) can be hierarchical, so we can

> > have TSC-H

> and TSC-L, to match MDSC-H and MDSC-L.

> >

> > 2) The -02 version does not clearly define "A higher level system",

> > which is

> needed.

> >

> > 3) It is needed to clearly describe which of these components are

> > slice-aware

> and which are not.

> >

> >

> >

> > Thanks,

> >

> > - Xufeng

> >

> >

> >

> > On Tue, May 12, 2020 at 12:03 PM Luis M. Contreras

> <[e-mail]<mailto:[e-mail]>> wrote:

> >

> > (My apologies for the mail before, with the Confidential Notes by default.

> Please, ignore it. Repeating mail here).

> >

> >

> >

> > Hi Eric, all,

> >

> >

> >

> > Thanks for addressing this point.

> >

> >

> >

> > I have some few comments and suggestions.

> >

> >

> >

> > 1/ Regarding the figure, I think that as it is now mix things a bit.

> > In my personal

> subjective view I think it provides a view of how to fit TSC into ACTN

> rather in the other way around. My suggestion would be to depict the

> comparison as in the figure below (ignore the service model naming by

> now for this point). I think it becomes more clear, but again this is subjective, of course.

> >

> >

> >

> > 2/ Regarding the service model naming. I have revisited RFC8309 and

> > I think

> that the proper mapping to service models in the case of TSC would be

> the one proposed in the figure below on the left-hand side (please,

> check Figure 3 in RFC8309).

> >

> >

> >

> > Here is the figure I propose.

> >

> >

> >

> > Service Model in                                              ACTN

> >

> >  Transport Slice        Transport Slice Framework           Terminology

> >

> > Framework [RFC8309]                                         and Concepts

> >

> > -------------          -------------------------           ------------

> >

> >                  +------------------------------------+

> >

> >                  |             Customer               |  |

> >

> >                  +------------------------------------+

> >

> >    Customer                        A                     |

> >

> >    Service                         |

> >

> >    Model                           V                     |

> >

> >                  +------------------------------------+

> >

> >                  |      A highter level system        |  |   +-----+

> >

> >                  |(e.g e2e network slice orchestrator)| ===> | CNC |

> >

> >                  +------------------------------------+  |   +-----+

> >

> >    Service                         A                            A

> >

> >    Delivery                        | TSC NBI             |      | CMI

> >

> >    Model                           V                            V (LxSM)

> >

> >                  +------------------------------------+  |   +-------+

> >

> >                  |      Transport Slice Controller    | ===> | MDSC-H|

> >

> >                  +------------------------------------+  |   +-------+

> >

> >                                    A                            A

> >

> >                                    |                     |      |

> >

> >                                    |                            V

> >

> >    Network                         |                     |   +-------+

> >

> >    Configuration                   | TSC SBI                 |MDSC-L |

> >

> >    Model                           |                     |   +-------+

> >

> >                                   |                            A

> >

> >                                    |                     |      | MPI

> >

> >                                    V                            V (LxNM)

> >

> >                  +------------------------------------+  |   +-----+

> >

> >                  |        Network Controller(s)       | ===> | PNC |

> >

> >                  +------------------------------------+  |   +-----+

> >

> >

> >

> >

> >

> > The text you propose depends totally on the final figure and the

> > consideration

> of the service models that apply, so I would prefer to discuss first

> about the figure and once it is closed to discuss the text, to ensure consistency.

> >

> >

> >

> > Thanks

> >

> >

> >

> > Luis

> >

> >

> >

> > El mar., 12 may. 2020 a las 16:03, Eric Gray

> (<eric.gray=[e-mail]<mailto:eric.gray=[e-mail]>>) escribió:

> >

> > John and I had a discussion with Dhruv about some ambiguity in the

> relationship between some of the entities in the ACTN architecture and

> the related concepts in our NS-DT work in the Framework draft.

> >

> >

> >

> > The upshot is that there is a bit of a slushy relationship between

> > some of the

> terms that we (the NS design team) have defined (see the definitions

> draft) and loosely corresponding concepts defined in ACTN.  In

> particular, there are currently issues with the positioning of the TSC

> as directly analogous to MDSC, impacting interfaces between these logical components as well.

> >

> >

> >

> > After a few iterations in discussion, Dhruv, John and I agreed to

> > propose a

> replacement figure and text as shown immediately below.

> >

> >

> >

> >        +------------------------------------+

> >

> >        |             Customer               |  |

> >

> >        +------------------------------------+

> >

> >                          A                     |     ACTN

> >

> >                          |                        Terminology

> >

> >                          V                     |  and Concepts

> >

> >        +------------------------------------+

> >

> >        |      A highter level system        |  |   +-----+

> >

> >        |(e.g e2e network slice orchestrator)| ===> | CNC |

> >

> >        +------------------------------------+  |   +-----+

> >

> >                          A                            A

> >

> >                          | TSC NBI             |      | CMI

> >

> >                          V                            V (LxSM)

> >

> >        +------------------------------------+  |   +-------+

> >

> >        |      Transport Slice Controller    | ===> | MDSC-H|

> >

> >        +------------------------------------+  |   +-------+

> >

> >                          A                            A

> >

> >                          | TSC SBI (LxNM)      |      |

> >

> >                          V                            V

> >

> >        +------------------------------------+  |   +-------+

> >

> >        |        Network Controller(s)       | ===> |MDSC-L |

> >

> >        +------------------------------------+  |   +-------+

> >

> >                                                       A

> >

> >                 Terminology/Concepts           |      | MPI

> >

> >                Used in this Document                  V

> >

> >                                                |   +-----+

> >

> >                                                    | PNC |

> >

> >                                                |   +-----+

> >

> >

> >

> >

> >

> > We would also add further clarifying text along the lines of:

> >

> >

> >

> > The TS-NBI would be at the same level as the customer service models

> > (LxSM),

> except that it uses a technology agnostic service model where as LxSM

> does not.

> >

> >

> >

> > We add hierarchy to the MDSC concept in this figure so that the

> > mapping with

> LxNM might be easier to see.  But the TSC could also directly interact

> with multiple domain controllers in which case we have a single MDSC.

> >

> >

> >

> > If nobody objects, or has additional input they would like to

> > provide, we would

> like to go ahead and make this change to the Framework draft.

> >

> >

> >

> > If necessary, we can discuss this at the meeting on Thursday (14 May).

> >

> >

> >

> > --

> >

> > Eric

> >

> > --

> > Teas-ns-dt mailing list

> > [e-mail]<mailto:[e-mail]>

> > https://protect2.fireeye.com/v1/url?k=1eaec4cb-400e2455-1eae8450-86e

> > e86bd5107-f483cc5da58c33bd&q=1&e=54580c1a-e1c9-4821-ab55-4d147d30d06

> > a&u=https%3A%2F%2Furldefense.com%2Fv3%2F__https%3A%2F%2Fwww.ietf.org

> > %2Fmailman%2Flistinfo%2Fteas

> > -ns-dt__;!!NEt6yMaO-

> gk!VAg0Vz27PX9nHj1eNBVGWt0LwRfVXuZn7V_O62U0sL-2uRR

> > uv_UFfyXBGARF8hg$

> >

> >

> >

> >

> > --

> >

> > ___________________________________________

> >

> > Luis M. Contreras

> >

> > [e-mail]<mailto:[e-mail]>

> >

> > [e-mail]<mailto:[e-mail]>

> >

> > Global CTIO unit / Telefonica

> >

> > --

> > Teas-ns-dt mailing list

> > [e-mail]<mailto:[e-mail]>

> > https://protect2.fireeye.com/v1/url?k=7fc5cb69-21652bf7-7fc58bf2-86e

> > e86bd5107-bcf121b69ec9b593&q=1&e=54580c1a-e1c9-4821-ab55-4d147d30d06

> > a&u=https%3A%2F%2Furldefense.com%2Fv3%2F__https%3A%2F%2Fwww.ietf.org

> > %2Fmailman%2Flistinfo%2Fteas

> > -ns-dt__;!!NEt6yMaO-

> gk!VAg0Vz27PX9nHj1eNBVGWt0LwRfVXuZn7V_O62U0sL-2uRR

> > uv_UFfyXBGARF8hg$

> >

> > --

> > Teas-ns-dt mailing list

> > [e-mail]<mailto:[e-mail]>

> > https://protect2.fireeye.com/v1/url?k=78f842d7-2658a249-78f8024c-86e

> > e86bd5107-764e815231c9d43f&q=1&e=54580c1a-e1c9-4821-ab55-4d147d30d06

> > a&u=https%3A%2F%2Furldefense.com%2Fv3%2F__https%3A%2F%2Fwww.ietf.org

> > %2Fmailman%2Flistinfo%2Fteas

> > -ns-dt__;!!NEt6yMaO-

> gk!VAg0Vz27PX9nHj1eNBVGWt0LwRfVXuZn7V_O62U0sL-2uRR

> > uv_UFfyXBGARF8hg$

>

> --

> Teas-ns-dt mailing list

> [e-mail]<mailto:[e-mail]>

> https://protect2.fireeye.com/v1/url?k=153b4b4d-4b9babd3-153b0bd6-86ee8

> 6bd5107-4b38ef80511ff65e&q=1&e=54580c1a-e1c9-4821-ab55-4d147d30d06a&u=

> https%3A%2F%2Furldefense.com%2Fv3%2F__https%3A%2F%2Fwww.ietf.org%2Fmai

> lman%2Flistinfo%2Fteas-ns-

> dt__;!!NEt6yMaO-gk!VAg0Vz27PX9nHj1eNBVGWt0LwRfVXuZn7V_O62U0sL-

> 2uRRuv_UFfyXBGARF8hg$