nmop

Summary

The first nmop session covered several key topics, including document status updates, cooperation with other SDOs, and presentations on the terminology draft, CMAP concept requirements, message broker, incident management YANG model, and operator outage experiences. The session concluded with a discussion on knowledge graphs and their potential applications in network management.

Key Discussion Points

Document Status: Updates were provided on adopted and near-complete documents, including the Network Anomalies Lifecycle, Semantic, and Technology drafts. The C-map concept requirements and use cases draft is a current focus.
SDO Cooperation: Collaboration efforts with IOWN, 3GPP, ETSI TC Data, and Broadband Forum were discussed. Specific areas of focus included network digital twins, network topology data, data governance, and time stamping in YANG Push.
Terminology Draft: Progress was reported on the terminology draft, including the addition of context terminology and removal of system-related definitions. Reviewer feedback is being addressed, and the working group is encouraged to continue reviewing and commenting.
CMAP Concept Requirements and Use Cases: Open issues were discussed, including terminology (multi-layer topology), multiple viewpoints, partitioning, interface/termination point definitions, the network design use case, use case grouping, passive topology requirements, and a list of models for CMAP navigation.
Message Broker: The session presented a draft for an accessible YANG model for telemetry notification, aiming to define the schema and data exchange between data collection components and the message broker. The focus is on a message envelope to add metadata.
Incident Management YANG Model: An update was provided on the YANG data model for network incident management, with a focus on terminology alignment (especially related to root cause), and clarification of relationships between problems, incidents, and customer incidents.
Operator Outage Experiences: DT and Swisscom shared outage experiences, highlighting challenges in monitoring ECMP paths and detecting blackholing scenarios. Specific incidents related to maintenance windows and router processor failures were discussed.
Knowledge Graphs: A discussion was initiated on the potential of knowledge graphs in network management, particularly for linking information and improving incident analysis.

Decisions and Action Items

CMAP - Multi-Layer Topology: Authors will keep the multi-layer topology terminology and add a clarification explaining that multiple network layers can be grouped into a single topology layer, and the relation to RFC 345's topology hierarchy.
CMAP - Multiple Viewpoints: Authors will add a requirement for different viewpoints to be presented via the APIs.
CMAP - Partitioning: Authors will add a requirement that partitioning should be supported by the model.
CMAP - Interface and Termination Points: The CMAP draft will refer to RFC 8345 for definitions of termination points and interfaces, quoting the relevant sections. Nigel agreed to provide material from another action item related to terminologies to the mailing list.
CMAP - Network Design Use Case: Discussion on the network design use case will continue on the mailing list among Brad, Benoit, and Sheriff.
CMAP - Use Cases Grouping: The CMAP draft will focus on the parts of use cases relevant to the Simap API and review if there is enough info for requirements
CMAP - Ivy Passive Topology: The group will double check on the mailing list if everyone's ok to leave passive topology to be optional and to associate and navigate from the C-map to the passive topology.
Incident Management - Root Cause: The incident management YANG model will re-evaluate the root cause terminology based on the discussions and seek a more appropriate term.
Message Broker - Andy: Authors of the message broker YANG model will reach out to Andy to review the proposed YANG model.
Knowledge Graphs: An interim meeting will be scheduled to further discuss the potential of knowledge graphs in network management.

Next Steps

Adrian and Rob will sit down tomorrow to resolve remaining issues with the terminology draft.
Adrian will post a new revision of the terminology draft soon, incorporating agreed-upon changes.
The working group is encouraged to continue reviewing and commenting on the terminology draft.
Authors of the C-map draft will address the agreed-upon actions and update the draft accordingly.
The incident management YANG model will be revised to address the concerns raised regarding the root cause terminology.
An interim meeting will be scheduled to discuss knowledge graphs in more detail.

Session Date/Time: 18 Mar 2025 08:30

nmop

Summary

The NMOP working group meeting covered several topics including hackathon results on Yang Push Publisher implementations, C-map for SRV6 and topology linking to external data, anomaly detection integration updates, and the transformation of Yang configuration instance data to knowledge graphs. A presentation was given on HTTPS notifications using Kafka and bandwidth analysis.

Key Discussion Points

Yang Push Publisher Hackathon Results (Thomas):
- Focus on notification envelope and system notification capability discovery.
- Testing extended to broadband access (OLT).
- Website created for implementers with test results and configurations.
- First CBO named identifier implementation tested successfully.
- Lack of DTLS/TLS implementations.
C-map for SRV6 and Linking Topology to External Data (Sheriff):
- Modeling of multi-layer networks using C-map.
- Linking external data (configuration, metrics) to the C-map model using templates.
- Two approaches for augmenting the C-map: direct augmentation (approach one) and an intermediate layer (approach two).
- Pros and cons of each approach discussed regarding scalability and complexity.
- Added SRV6 on top of ISIS to the C-map model.
Anomaly Detection Integration Update (Vincenzo):
- Updates on the Network Anomaly Lifecycle document and Semantic Metadata Annotation draft.
- Focused on young format and fixing key errors.
- Addition of a global concern score at the relevant state level.
- Added URI to link anomaly visualization to the anomaly object.
- Antagonist (open-source label store) being used to validate drafts.
Yang Configuration Instance Data to Knowledge Graph (Michael):
- Experiment with transforming Yang instance data into RDF (Resource Description Framework) using RML (RDF Mapping Language).
- Using Sparkle to query the RDF data.
- Conclusion that this was not a good problem approach due to scalability issues.
- Poor tooling in the RDF and property graph worlds.
- Suggesting the use of streaming approaches with XML tools instead.
HTTPS Notifications and Bandwidth Analysis (Students):
- Implemented HTTPs notifications using Kafka and performed bandwidth analysis of different encoding formats (CBO, XML, JSON).
- CBO performed better at lower bandwidths.
- Proposed draft to extend the existing HTTPs notifications draft to support CBO encoding.
- Lack of tooling for CBO in Yang libraries was an issue.

Decisions and Action Items

Adrian: To make minor edits to the terminology draft and continue detailed survey of terminology usage in existing NMOP drafts.
Sheriff: Engage with the NMOP working group to get feedback on approach one vs. approach two for linking external data to C-map, with the goal of proceeding with one approach.
Michael: To address scalability issues related to RDF graph generation for IETF 123 Hackathon in Madrid.
Lionel/Nacho/Michael: Explore possible joint discussion on Airmen deployment
Michael: Share experiences and feedback on using RML for YANG instance data
Students: (NITK Students) To continue work on their implementation, and explore contributing to open source Cibor and Gansson libraries to add Cibor support.

Next Steps

Continue hackathon activities to validate and refine the drafts.
Evaluate and address feedback on the proposed approaches.
Further discussion on knowledge graph approaches.
Continue development of running code to accelerate specifications.