Should network inventory augment RFC8345 network or define a new root?

[italobusi]

There are two options where to place the network inventory information

Option 1: within the network. as suggested in RFC8345

augment nw:networks/nw:network:
   <..................>

Option 2: under a new root:

+--ro network-inventory
   <..................>

Option 1 is more aligned with the text in RFC8345

Option 2 simplifies the operations for the inventory application when retrieving only the inventory information without the need to know the network-id assigned by the server to the network instance reporting the network topology

[italobusi]

Network Inventory weekly call (October 13, 2021)

Jeff: today proprietary APIs are following option 2. However, we need also to be able to correlate the topology information with the inventory information.

Sergio: to be checked if using filters would solve the issue

Agreement to follow option 2 for the -00 and to keep the option open for further discussion/investigation

[italobusi]

• [ ] @agva123 @YuChaode : could you please check (e.g., asking the netconf WG) if this is possible with RESTCONF protocol?

[YuChaode]

I discussed with the author of RESTCONF and found it's hard to use exsisting parameter to do the filtering if we define our inventory model under the network model. And Dr. Zheng from China Unicom also suggests us to have a new root.

[italobusi]

It has been agreed to augment RFC8345.

[italobusi]

2023-02-15 Network Inventory weekly call

It may worthwhile to add some text in the document to clarify the rationale for this choice

• [ ] @YuChaode provide some initial text proposal

See: https://github.com/italobusi/ietf-network-inventory/issues/60#issuecomment-1432999267

[italobusi]

See also comments from Oscar in #75

[italobusi]

Closed in draft-y3bp-ivy-network-inventory-yang-00