Session Date/Time: 08 Oct 2024 10:00

NMRG

Summary

The NMRG interim meeting focused on updating the group's research agenda, discussing four key topics: AI for Network Management, Data-Driven Networking, Network Digital Twin (NDT), and Green Networking/Sustainability. The Chairs presented drafted text for each area, inviting feedback from participants to refine research questions, identify priorities, and determine the scope of NMRG's future work. A significant point of discussion was the future of Intent-Based Networking (IBN) within the group's research agenda, with a call for continued discussion on the mailing list. The aim is to finalize an updated research agenda by the next IETF meeting.

Key Discussion Points

The Chairs, Jerome Fritsch and Laurentius van der Velde, opened the meeting by reiterating the IETF Notewell, privacy, and code of conduct policies. They emphasized NMRG's role as a research group focused on long-term research issues, publishing informational or experimental RFCs, rather than developing standards. The primary goal of the meeting was to discuss the group's research agenda, which has been under development for several months.

Research Agenda Introduction

The Chairs outlined the process of updating the research agenda, which aims to provide a clear roadmap for the group's activities.
Emphasis was placed on focusing on fundamental research questions and building understanding in research areas, rather than starting with solutions or specific standards.
The need to prioritize topics was highlighted due to limited workforce and the importance of maintaining focus.
Participants were encouraged to consider ongoing work in other IETF WGs and external research venues.
The goal is to reach a refined research agenda for consensus at the next IETF meeting.

AI for Network Management

Discussion centered around a draft document outlining objectives and potential work items.
Key research challenges identified included:
- Deployment issues and interoperability of AI-based solutions with existing network management technologies.
- Generalization of AI models across different network environments.
- Explainability, trustworthiness, robustness, and data lineage for AI in network management.
The group aims to avoid developing reference architectures or focusing on highly specific use cases, instead seeking common research findings and questions.
A related area of interest is how networking capabilities can be leveraged to enhance AI-based services and applications (the "inverse" relationship).
Laurentius van der Velde suggested investigating the transformative impact of pervasive AI on how network management and operations are performed, considering potential changes in functionalities and interactions.
Luis M. Contreras proposed exploring the interplay of AI capabilities deployed at different network levels (e.g., transport networks, IP level) and their coordination.

Data-Driven Networking

This topic was presented as being beyond the sole scope of AI, addressing general data problems in network management.
Key research questions include:
- Addressing data quality problems for network management applications (e.g., network traffic, routing policies).
- Developing quality metrics, "quality by design" principles, and methods for assessing data quality.
- Proposal: Create a rating or assessment framework for commonly used datasets in network management research.
- Improving data usability: understanding dataset limitations, exploring new data models/formats, and more efficient data collection.
Laurentius van der Velde emphasized data's central and transversal role in network management and automation. He suggested exploring synergies with MAPRG (Measurement and Analysis for Protocols Research Group) and potentially holding a joint meeting to discuss data questions.
Luis M. Contreras highlighted the complexity of correlating data from different network layers and segments (e.g., latency across radio, transport, virtualization) to build an end-to-end view for customers.

Network Digital Twin (NDT)

Discussion referred to a draft outlining potential topics, though acknowledging it might not reflect the very latest discussions.
Areas of interest included:
- Enabling technologies for NDTs (e.g., AI, simulation, virtualization).
- Characterizing NDTs: establishing criteria (e.g., real-time vs. non-real-time, network layer scope, underlying technologies) rather than seeking a single definition.
- Composing and comparing multiple NDTs, including qualitative and quantitative comparisons.
- Deployment and interoperability challenges for NDT solutions.
Laurentius van der Velde suggested investigating the impact of NDTs on how network management is performed, potentially viewing NDTs as the next generation of Network Management Systems (NMS). He also emphasized exploring "architectures in plural" for NDTs rather than aiming for a single reference architecture, and connecting NDT studies with data quality and trustworthiness.
Chang Liu agreed on the importance of enabling technologies, use cases, validation scenarios, and characterization criteria. He provided an update on the "Network Digital Twin Concept and Architecture" draft, stating that a new version has been uploaded to the data tracker addressing most issues and is very close to completion. He offered to provide a detailed update at the next NMRG meeting.
The Chairs noted that validation scenarios and implementations (e.g., via hackathons) could be valuable for NDT, similar to past IBN efforts.

Green Networking / Sustainability

The topic's inclusion was largely supported by a recent survey of NMRG participants, indicating its importance.
A current draft on "Green Networking Challenges" is under IESG review.
Discussion points included:
- Broadening the scope beyond just energy considerations to wider environmental impacts and UN Sustainable Development Goals (SDGs).
- Focus on metrics (e.g., energy, carbon), their calculation, approximation, and integration into network management.
- Assessing trade-offs between energy efficiency gains and potential performance loss.
The group acknowledged the transversal nature of the topic and the need to carefully scope NMRG's role given other IETF activities (e.g., Green WG, IESG programs).
Luis M. Contreras suggested focusing on research areas not directly covered by the Green WG, such as carbon emissions, carbon-aware traffic engineering, and the lifecycle of network equipment.
Laurentius van der Velde cautioned against overly broad scope ("greenwashing") and advocated for identifying very specific sub-areas where NMRG can make a relevant research contribution. He suggested NMRG could focus on a network-wide view for coordinating and correlating energy-related information, and the role of network management in interfacing between network infrastructure and broader sustainability goals of operators/consumers.
Jerome Fritsch shared survey highlights: top interest areas are measurement/telemetry (beyond protocols/formats) and the correlation between energy efficiency and functional performance. Cross-layer optimization also received support.
The Chairs invited participants with interest in green networking to propose specific initiatives (e.g., new documents, discussions with other communities) to move the work forward.

Intent-Based Networking (IBN) Update

The scheduled presentation by Paul E. Paul was not given due to his absence.
Discussion proceeded on the "IBN Use Cases and Practices" document, a joint draft merging previous use cases and identifying commonalities and lessons learned.
The core question for the group is whether the research phase of IBN within NMRG is largely complete.
The Chairs noted that IBN has been a long-standing topic in NMRG with significant positive outcomes, fulfilling most of its initial work plan. The current use cases document is considered an achievement.
Laurentius van der Velde questioned whether IBN should remain a standalone topic in the updated research agenda or if it's considered largely achieved, with future presentations always welcome but not a core focus.
Luis M. Contreras requested that the discussion regarding the continuation of IBN within NMRG be formally brought to the mailing list to gather broader consensus, particularly from those not present at the interim meeting (e.g., Paul).

Decisions and Action Items

The discussion on the NMRG research agenda, including the future of IBN as a standalone topic, will continue on the NMRG mailing list.
Chairs will provide specific questions to guide this mailing list discussion.
Chang Liu will provide a detailed update on the "Network Digital Twin Concept and Architecture" draft at the next NMRG meeting.
Chairs: Update the research agenda draft based on feedback received during this meeting.
All participants: Review the updated research agenda draft and contribute feedback on the mailing list, particularly on the focus areas and the future of IBN.
Participants interested in Green Networking: Propose specific initiatives for NMRG to undertake in this area.

Next Steps

The Chairs will refine the research agenda document based on the feedback from this interim meeting.
A formal discussion will be initiated on the NMRG mailing list regarding the updated research agenda and, specifically, the role and scope of Intent-Based Networking (IBN) in future NMRG activities.
The objective is to finalize the NMRG research agenda by the next IETF meeting.