Allotted Resource Model Review Comments

 #1: @Michela Bevilacqua I report here some of the comments previously provided in Allotted Resource Revised to be further discussed.

 I would reccomend to discuss a possible UC for this model to understand how it is expected to be used and initiated these new. classes Without a context it is quite impossible to understand what/who/when/how they should be used. Who is the actor going to define the ANFD ?  When a ANFD and ANFI is going to be created ?

 @Andy Mayer : The vCPE Use Case from the Amsterdam release includes a discussion on Allotted Resources. See: Use Case: Residential Broadband vCPE (Approved)

@Michela Bevilacqua : THanks Andy

Open

I have captured here the Allotted Resource model discussed in the context of the vCPE UC on the above link.

I see here a problem. The vCP example uses the old ANF resource model proposal and not the new ANF Revised model.

I now doubt which is the final model problem we should consider. in addition, if the vCPE UC implemented an old model, I´m not sure now if we have ONAP sw impacts if we want to introduce a new model.

In addition, can you help us understanding the introduction of this new class even in the context of the vCPE example, When a ANFD and ANFI is going to be created ?

What does "TunnelXConn" represent in the example?

 AnfInstance vs ServiceCompositeInstance

 #2: @Michela Bevilacqua : I report here some of the comments previously provided in Allotted Resource Revised to be further discussed.

What ANF is going to add respect to the ServiceCompositeInstance ?

Anfd and AnfInstance

#3: @Michela Bevilacqua Please add the inheritance tree for the new classes.

 Anfd and AnfInstance

 #4:  @Michela Bevilacqua  Which module, the new classes, Anfd and AnfInstance should belong to ? Vnf::ObjectClasses as qualified name refer to the intention to add them in the VNF module. I do not think this new class is constrainted only to VNF so I would reccomend to evaluate some different options:

1) include it in the Common Module

2) introduce a new Nf module where collect some of the Common classes as Nfx + Anfx

3) introduce a new dedicated Anf module

 Anfd and AnfInstance diagram

 #5 @Michela Bevilacqua : Note about Pnf  is not clear. It states: "Pnf cannot run on a Vnf or a Anf". Pnfs cannot run on Pnfs, etc" but the new relation introduce to the Anf to run on something. The new relation therefore automatically excludes a Pnf as well as a Vnf running on something. The question unanswered is really more about the scope of the Anf.

Can an Anf run on an Anf  ?

Can an Anf run on a Vnf ?

Can an Anf run on a Pnf ?

I do not see other options. Right ?

Anfd and AnfInstance diagram

#6 @Michela Bevilacqua  Note about AnfInstanceRunson.... is not clear.

What does it mean it represent the provided by ?

The indication about the service instance is not clear as this relation is about Nf instances. Can you clarify ?

 AnfInstance

#7: @Michela Bevilacqua The definition of the AnfInstance is ambiguous.

It does not describe any real characteristic of this new class byt just say it is provided by some service instance. 

Actually there is no new relation between the service instance and the anfInstance more than any other nfInstance.  We are lacking of a differentiator for this NF type. Can you elaborate further ?

AnfInstance

#8: @Michela Bevilacqua  selfLink attribute definition is not clear.

We are speaking of a link to external information provided by another ONAP component , a controller.

This attribute does not fit with an information model but more with a realization model.  We agreed from the beginning that our model should abstract from the realization model of each ONAP components.

What is the data/information we think it is important to model ?  

AnfInstance

#9: @Michela Bevilacqua orchestrationStatus:

This attribute is not clear. what is the relation between an orchestrator and this class ?

Isnt´t it enought the operationalStatus ?

Nfd and NfInstance, Anfd and AnfInstance

#10: @Michela Bevilacqua I report here some of the comments previously provided in Allotted Resource Revised to be further discussed.

A definition of the two new proposed relationships  is missing : ANFDRunsOnSharableNFDesc and ANFInstanceRunsOnSharableNFInstance

Nfd and NfInstance, Anfd and AnfInstance

#11: @Michela Bevilacqua I report here some of the comments previously provided in Allotted Resource Revised to be further discussed.

we should avoid a recursive relationship reviewing the modelling proposal. 

AnfInstance

#12: @Michela Bevilacqua : AnfInstance introduces a number of very detailed attributes that in my view do not fit well with an abstract concept of ANF. I would suggest to consider if we should prefer a more generic approach as we did for network slicing paramenter introducing a generic parameter profile concept