Liaison statement
Response to BBF Liaison "New project in BBF entitled "network resource model (NRM)” - questions answered

Dear Mr Lavoie,

Based on several informal discussions since our prior liaison response
(https://datatracker.ietf.org/liaison/1901/), we wished to provide you with a
more detailed set of information and answers to your questions from your
original liaison on the subject topic
(https://datatracker.ietf.org/liaison/1895/ |  BBF LIAISE-644).

Before getting into details, we wish to inform you that, while network
inventory is owned by the IVY working group, RFC 8345 (IETF topology) was
produced by a closed working group of the routing area (I2RS) and that the
recently created NMOP working group
(https://datatracker.ietf.org/group/nmop/about/) is working on issues related
to the deployment/usage of YANG topology modules. NMOP is the right venue to
discuss evolutions of RFC 8345. Regarding your liaison questions related to IVY
design decisions, please find our answers in line below.

If you have any further questions, please don't hesitate to ask.  As always,
the most expedient form of information exchange is directly via the IVY WG
email list.  If more interactive discussion is required, we are happy to
arrange one or more IETF IVY interim meeting calls to facilitate such upon
request.

Thank you for your interest and we look forward to your participation on the
IVY WG email list and meetings from those in BBF interested in the IVY work.
Please see the links below to help facilitate IVY WG participation.

Regards,

Daniele Ceccarelli and Quifang Ma

IETF IVY WG Co-Chairs

References:

·         IETF IVY WG Page (Datatracker) - including information for emai list
subscription, and meeting schedule https://datatracker.ietf.org/wg/ivy/about/

Answer details to BBF Liaison "Response to BBF Liaison "Network Resource Model
(NRM)""

·         What is the rationale behind the decision from IETF perspective for
creating a new network inventory data model in IVY WG instead of augmenting RFC
8345?

[IVY] From an IETF perspective, the rationale behind the decision for creating
a new network inventory data model in IVY WG instead of augmenting RFC 8345 is
that the WG would like a consistent view of a list of equipment in the network
which is separate, independent and standalone from network topology content
despite some possible similarities.   Inventory and topology are different
concepts.  For example, many things in topology are not useful for inventory,
such as logical network topology information. Conversely, inventory may cover
some cases that are not suitable to build on top of topology given it is
focused on what is currently deployed in the network. e.g., historical
inventory or planning inventory.  The WG does not wish to mix these concepts
and information together.  Network operators can benefit from such visibility
to keep track of what devices are deployed in the network including a variety
of information like components and software/hardware versions. Recent
discussion also reveals that network inventory may need cover not only what is
already deployed in the network but everything that is “owned” by the network
operator, including e.g., spare equipment that is planned to be deployed. The
WG may also identify other requirements in the future for network inventory
that are inappropriate to build on top of network topology. For reasons
outlined above, the IVY WG consensus is to create a new network inventory data
model instead of augmenting RFC 8345.”

·         RFC 8528 defines a YANG Schema Mount. Has IETF considered to use such
schema for a network inventory?

·         What are the reasons why it was not considered or why it was not
selected as a solution for network inventory?

[IVY] It has been considered but not adopted.  It was not selected as it was
considered a too complex solution for our purpose. Here are some details:

1.       decoupling the NBI from SBI would allow some filtering/abstraction not
to expose information which is needed by the domain controller but not by the
hierarchical controller;

2.       one objective of the work is to support a standard model for network
inventory (at the controller’s NBI) even in cases where the device model (at
the controller’s SBI) is proprietary;

3.       another objective is to define a common network inventory model for
optical, IP and MW networks even if the reference standards at the SBI are
different.

·         What are the practices in IETF for supporting intent configuration
for network topologies using RFC8345?

[IVY]  RFC8345 supports intent configuration in RW containers/attributes, so
that network topologies are configurable by the client, e.g., in the case of
overlay topologies.

Section 4.4.10 (https://datatracker.ietf.org/doc/html/rfc8345#section-4.4.10)
states the support of client-configured network topologies.

·         What is the consideration from IETF IVY WG to support configuration
for network inventory.

[IVY]  This is being discussed. We started with all the containers/attributes
being read only but depending on the use cases to be supported we might need to
have RW capabilities. Given that this topic is still under discussion, would it
be possible to share more details about the requirements to support
configuration of network inventory?  The best way to accomplish this is for
interested individuals to comment on the IETF IVY WG email list.

·         Currently, all the data nodes in the IVY network inventory DM are
defined as “read-write”. Is this network inventory a direct target for network
or service build intents from the NB layer (CloudCO DO or OSS)?

[IVY]  The IVY core model is not planning to cover services. The discussion
whether to provide RW or RO capabilities is still ongoing and will highly
depend on the list of use cases that the working group will decide to support. 
The most recent version of the IVY network inventory data model has defined
some data nodes as “read-only” while others as “read-write”.

·         What is the approach and rationale for modeling the relationship
between topology and inventory elements?

[IVY]  One of the items in the IVY charter is covering this item: “Mapping and
correlation semantics: Correlating the inventory with existing IETF models
e.g., topology, service attachment points (SAP), etc.”  A yang data model to
build the mapping between topology and inventory elements is planned to be
delivered by IVY WG.

·         What is the plan for IETF to finalize the network inventory model and
publish it? The state of the IETF network inventory DM may impact the release
of the BBF inventory model should BBF choose to augment the IETF DM and build a
new BBF inventory DM.

[IVY]  The core model is already a working group item, meaning that it is an
official IETF document that the working group is working on. According the IVY
WG charter, the network inventory model focuses a narrow scope with the
estimation to be able to run the working grouping last call by the end of 2024.
Unfortunately, it is extremely hard to predict how much time the draft will
take then to pass IETF last call and become RFC. Once the core model will be
delivered, we will build new use cases and extensions as augmentation to it.