search

ietf-ivy-wg / network-inventory-topology

Other
0 stars 0 forks source link

Italo's comments on the YANG model improvements during WG adoption #2

Open lana-wu opened 5 hours ago

lana-wu commented 5 hours ago
  1. I have seen you have added appendix A but it is not very clear what is the purpose of this Appendix (at the first glance it appears a copy of the YANG model which we developed in the CCAMP WG I-D): do we really need this appendix?
  2. What is the purpose of the node-name?

If it is intended to report the system-hostname, let's consider whether this attribute belongs to this model or to the network-element in the inventory model. If it belongs to this model, I would suggest to align the attribute name and description with the definition of system-hostname

If it is intended to be a generic name attribute, we may remove it as proposed by Olga during the meeting today

  1. What is the purpose of the system-mount-point empty container?

If the intention is to mount technology-specific attributes, I am wondering whether using a "multi-inheritance" approach, as described in section 4.1 of RFC8345, could be used to specify technology-specific attributes and navigation from topology to inventory within a single topology instance.

  1. What is the purpose of the tp-name?

I think we can remove it as proposed by Olga during the meeting today

  1. Why the tp-name is RO while the node-name is RW?
  2. What is the purpose of the leaf-list interface-name?

If the intention is to navigate from topology to interface model, what about using the ietf-tp-interface-reference-topology model in draft-ietf-ccamp-if-ref-topo-yang?

This I-D is a CCAMP WG draft because we have spun it off from the MW topology model after thinking that the requirement to navigate from topology to interface is technology-agnostic.

It has presented to Netmod WG in IETF 118 and still looking for an home WG where to be progressed: minutes-118-netmod-202311071200-00.md (ietf.org)https://datatracker.ietf.org/meeting/118/materials/minutes-118-netmod-202311071200-00%3Chttps//datatracker.ietf.org/meeting/118/materials/minutes-118-netmod-202311071200-00%3E)

  1. I would suggest to provide some text to address the question from Aihua during the meeting today.

Italo

lana-wu commented 5 hours ago

Candidate proposal:

1) I have seen you have added appendix A but it is not very clear what is the purpose of this Appendix (at the first glance it appears a copy of the YANG model which we developed in the CCAMP WG I-D): do we really need this appendix?

[Bo Wu] Thank for catching this. The comparison conclusion should be added that the modeling approach in CCAMP WG I-D seems redundant and causes these duplicate nodes at each layer. Defining a new "network-types" can avoid this and can be flexibly combined with other layer topologies.

2) What is the purpose of the node-name?

If it is intended to report the system-hostname, let's consider whether this attribute belongs to this model or to the network-element in the inventory model. If it belongs to this model, I would suggest to align the attribute name and description with the definition of system-hostname

If it is intended to be a generic name attribute, we may remove it as proposed by Olga during the meeting today

[Bo Wu] Agree with your suggestion as Olga pointed out to remove this and consider moving "system-hostname" to base inventory model.

3) What is the purpose of the system-mount-point empty container?

If the intention is to mount technology-specific attributes, I am wondering whether using a "multi-inheritance" approach, as described in section 4.1 of RFC8345, could be used to specify technology-specific attributes and navigation from topology to inventory within a single topology instance.

[Bo Wu] This node is defined for the augment target for obtaining some general configuration information, for example, ACL policies in the enterprise campus networks.

4) What is the purpose of the tp-name?

I think we can remove it as proposed by Olga during the meeting today

[Bo Wu] OK.

5) Why the tp-name is RO while the node-name is RW? [Bo Wu] Good catch. Will correct this.

6) What is the purpose of the leaf-list interface-name?

If the intention is to navigate from topology to interface model, what about using the ietf-tp-interface-reference-topology model in draft-ietf-ccamp-if-ref-topo-yang?

This I-D is a CCAMP WG draft because we have spun it off from the MW topology model after thinking that the requirement to navigate from topology to interface is technology-agnostic.

It has presented to Netmod WG in IETF 118 and still looking for an home WG where to be progressed: minutes-118-netmod-202311071200-00.md (ietf.org)https://datatracker.ietf.org/meeting/118/materials/minutes-118-netmod-202311071200-00%3Chttps//datatracker.ietf.org/meeting/118/materials/minutes-118-netmod-202311071200-00%3E)

[Bo Wu] Thanks for information. Will take this into account.

7) I would suggest to provide some text to address the question from Aihua during the meeting today. [Bo Wu] Agree to add text to clarify mapping relationship between nodes and NEs, also with other links and tps.