[OPSAWG]Re: [NMOP] Re: [nmrg] Re: [OPSAREA@IETF126] IRTF/IETF OPS Transfer: Status & Exploring Collaboration Opportunities

["[e-mail]" <[e-mail]>]

Dear Chongfeng, Ricardo and all:

Glad to see that many people starting to pay attention to AI agents for network management. 
We submitted a draft on network management agents (NMA) to NMOP back in 2024, hope this can serve as reference material for discussions.

URL: https://datatracker.ietf.org/doc/draft-zhao-nmop-network-management-agent/

I agree that the capabilities of AI agents include perception, planning and reasoning, tool invocation and execution, plus learning and self-optimization, as Chongfei mentioned. 

Given that NMA is designed to enable a closed loop from intent to execution, in the latest version of the NMA functional architecture, we have mapped each NMA function to the five phases of the TMF autonomous closed loop, namely intent management, perception, analysis, decision-making and execution (as shown in the figure down below).

About the "pre‑execution validation layer" that @Ricardo mentioned. I fully agree that it's critical and it should be done during the decision-making phase. 
This involves using capabilities such as LLMs or digital twin systems to simulate and evaluate proposed solutions, and determine whether to execute based on the evaluation outcomes.


Cheers
Xing
 
From: Chongfeng Xie
Date: 2026-06-28 11:47
To: 【外部账号】April
CC: ops-area; opsawg; nmop; [e-mail]
Subject: [NMOP] Re: [nmrg] Re: [OPSAREA@IETF126] IRTF/IETF OPS Transfer: Status & Exploring Collaboration Opportunities

Hi Ricardo,
Thank you for raising the point of pre‑execution validation, which I think is essential for the deployment of AI Agent for network operation. When you mention the term 'layer,' are you referring to a layer of a specific architecture or framework?

Best regards
Chongfeng
 
From: 【外部账号】April
Date: 2026-06-25 11:33
To: Chongfeng Xie
CC: Thomas.Graf; ops-area; opsawg; nmop; [e-mail]; [e-mail]
Subject: Re: [nmrg] [NMOP] Re: [OPSAREA@IETF126] IRTF/IETF OPS Transfer: Status & Exploring Collaboration Opportunities
Dear Chongfeng and all,
Thank you, Chongfeng, for the detailed and practical overview of AI agent capabilities in network management. Your analysis of perception, planning, tool use, and the applicability of MCP and YANG is very helpful.
I would like to raise one additional aspect that I believe is critical but often overlooked: the need for a pre‑execution validation layer that assesses whether an AI agent’s proposed action is physically admissible before it is committed to the network. 

Best regards,
Ricardo 

On Jun 24, 2026, at 9:14 PM, Chongfeng Xie <[e-mail]> wrote:



Hi Thomas,

Thank you for raising this topic for discussion. For the questions about AI agents, I'd like to provide my understanding for your reference.  Network management is a specific application of AI, and the management functions can be implemented across various AI agents that interact with the different components of the network management system. 

Q1. Which AI agents capabilities are suitable for network management?   
   A:   AI agents have the following common capabilities, 

    - Perception: Processes multi-modal inputs (text, images, audio) and tracks context and state over time.

    - Planning & Reasoning: Breaks high-level goals into actionable sub-tasks, dynamically re-plans when obstacles arise, and evaluates trade-offs before acting.

    - Tool Use & Execution: Autonomously calls external tools—search engines, APIs, calendars, code interpreters, and databases—to take real action in the digital world (e.g., send emails, book flights, query data).

    - Learning & Self-Correction: Detects errors, reflects on outcomes, iterates on solutions, and utilizes both short-term and long-term memory to improve over time.

All the capabilities above are useful for network management. With the underground support of LLM models, the following functionalities can be realized by AI agents in the context of network management,
      
   - Intent-based auto-assignment: Automatically routes NOC tickets based on detected intent.

   - Intelligent root-cause analysis: Pinpoints fault origins with precision.

   - Actionable troubleshooting guidance:  Generates step-by-step plans and recommended resolutions.

   - AI-driven orchestration:  Enables expert-human collaboration and automated execution of task reassignments/suspensions.

   - Continuous network optimization:  Delivers ongoing tuning recommendations for wireless, core, and cloud-network resources across diverse scenarios.

   - Automated recovery verification:  Validates fault resolution by querying network and perceptual metrics via LLM-orchestrated API calls.
    
Q2. What is the impact of an AI network management process to YANG information, service, network and data modelling?
    A:   AI and legcay network management models belong to different domains,  the specific manangement modelling is aganostic to the AI network manangent process. In other words, AI system can process any YANG information, service, network and data modelling.  

Q3.  Is MCP (https://en.wikipedia.org/wiki/Model_Context_Protocol) a suitable protocol between AI agents and IETF network management data sources and system components?
     A:  Yes, as an open standard that connects AI models to external data and tools, MCP enables AI assistants to securely read files, query databases, and execute actions across different applications.  The likelihood of MCP being used between between AI agents and IETF network management data sources and system components is very high. 
In addition, other approaches can be used, such as APIs exposed by network controller, in this case, AI agents can call the APIs directly for network device configuration or data collection. 

Q4, How does an AI agent interact with Network Digital Twin, Intent-Based Networking, SIMAP, Knowledge Graphs and YANG data in Message Brokers?
  A:   It depends on the specific situation, there should be no one-size-fits-all solution.

The above feedbacks are based on our current practice.  As AI for network is still in its early stages, our understanding may not be entirely accurate. We welcome any differing opinions.

Best regards
Chongfeng

From: [e-mail]
Date: 2026-06-21 16:43
To: [e-mail]; [e-mail]; [e-mail]; [e-mail]
CC: mohamed..boucadair
Subject: [NMOP] Re: [OPSAREA@IETF126] IRTF/IETF OPS Transfer: Status & Exploring Collaboration Opportunities
Dear NMRG, NMOP, OPSAWG and OPSAREA,
 
First of all many thanks to Brad, Dan, Christopher, Alex, Qin and Bernard on their feedback. That’s a great start.
 
I have been following this thread closely https://mailarchive.ietf..org/arch/msg/nmop/gTPL1eBj5e3GMMt8qeUWmPP4oMg/ to understand wherever research and industry adoption comes closer together, wherever there is interest for having a joint discussion and on what particular topic.
 
What I observed so far is that we all share that AI in general and also AIOps research and adoption rate has dramatically increased. So far everybody was very supportive on having a joint discussion. I would be interested to have more feedback, especially also from those who think that it would not bring much benefit with reasons why.
 
From what I observed, having an initial discussion on the following existing working group document topics and how they intersect
 
Network Digital Twin (https://datatracker.ietf.org/doc/html/draft-irtf-nmrg-network-digital-twin-arch)
Intent-Based Networking (https://datatracker.ietf.org/doc/html/draft-irtf-nmrg-ibn-usecases)
SIMAP (https://datatracker.ietf..org/doc/html/draft-ietf-nmop-simap-concept)
YANG-Push to Message Broker integration (https://datatracker.ietf.org/doc/html/draft-ietf-nmop-yang-message-broker-integration)
 
and then having a discussions on the following raised questions perhaps:
 
Which AI agents capabilities are suitable for network management?
What is the impact of an AI network management process to YANG information, service, network and data modelling?
Is MCP (https://en.wikipedia.org/wiki/Model_Context_Protocol) a suitable protocol between AI agents and IETF network management data sources and system components?
How does an AI agent interact with Network Digital Twin, Intent-Based Networking, SIMAP, Knowledge Graphs and YANG data in Message Brokers?
 
and move from there into some proposed work addressing these areas. And last but not least maybe wrapping up around running code. How experiments could facilitate and exchange between both working groups. Alex brought up that ANRW is once a year and perhaps on one or two other occasion the two working group could continuously actively exchange every year could be an interesting proposal.
 
Would that be a potential interesting agenda for everybody?
Did I miss anything or shall we focus more on a particular item?
Any particular discussion structure you would fancy?
 
I also observed that authors authoring Knowledge Graph documents have voiced wherever Knowledge Graph could be a topic of interest. I would love to see more discussions from other areas on this in particular how this intersects with AIOps and AI assisted network management topics. I saw also that ontologies have been mentioned, which I personally besides semantics believe plays an important role with interacting with AI agents.
 
As I am human, and not an AI agent, the observations and suggestions might be unintentionally bias. So please feedback and comment.
 
Best wishes
Thomas
 
From: Graf Thomas, SCS-INI-NET-VNC-E2E 
Sent: Monday, June 1, 2026 7:16 AM
To: Nmop <[e-mail]>; [e-mail]
Cc: Mohamed Boucadair <[e-mail]>; nmrg-chairs <[e-mail]>; Nmop-chairs <[e-mail]>
Subject: FW: [OPSAREA@IETF126] IRTF/IETF OPS Transfer: Status & Exploring Collaboration Opportunities
 
Dear NMRG and NMOP,
 
Med approached me wherever I am willing to facilitate with the support of both working group chairs a discussion between NMRG and NMOP on common interests.
 
Looking at the list of current adopted documents
 
https://datatracker.ietf.org/group/nmrg/documents/
https://datatracker.ietf.org/wg/nmop/documents/
 
but also looking at AIOps related proposals https://github.com/IETF-OPS-AD/misc/blob/main/ai-submissions-ietf125.md, and the related past side meetings in the network management domain, I see that the lifecycle on research and industry adoption of model and data driven network management becomes accelerated and therefore research and industry adoption comes more and more closer together. Looking at the bluesheets of IETF 125, I see that we had 107 participants for NMRG and 115 for NMOP and 36 participants were attending both NMRG and NMOP.
 
Or in other words, I personally believe that both working groups have a lot in common but also a also few things which makes them very unique and interesting to facilitate such a discussion.
 
I might be wrong and therefore would like to get some feedback from both mailing lists first:
 
Would you agree that the lifecycle on research and industry adoption of model and data driven network management becomes accelerated and therefore research and industry adoption comes more and more closer together?
 
Is there an interest at the NMRG and NMOP communities to have an exchange on currently adopted and future proposed work to identify synergies and common interests in both working groups and discuss how they could be addressed better?
 
Do you have a particular topic, set of documents or interest already in mind?
 
Best wishes
Thomas
 
From: [e-mail] <[e-mail]>
Sent: Monday, May 4, 2026 10:21 AM
To: Graf Thomas, SCS-INI-NET-VNC-E2E <[e-mail]>; [e-mail]; Nmop-chairs <[e-mail]>
Cc: ops-ads <[e-mail]>; Dirk KUTSCHER <[e-mail]>
Subject: [OPSAREA@IETF126] IRTF/IETF OPS Transfer: Status & Exploring Collaboration Opportunities
 
Be aware: This is an external email.
 
Hi all, 
 
I had several discussions recently with many IETF participants about the relationship between some work done in the IRTF and IETF. Specifically, I heard several questions about the current mode of operation between NMRG and OPS and whether/how activity transfer is happening between the two communities. I do see a value in bringing some of these items from hallway discussion to our formal meetings :-)
 
This can be the opportunity to explore between collaboration between the communities, and not rely only on the leadership level. See, for example, the following excerpt form the NMRG Charter:
 
   “The IETF Operations and Management Area Directors are members of the NMRG
   mailing list and invited to NMRG meetings in order to ensure free flow of
   information in both directions, and to avoid duplication of work with the
   various IETF working groups..”
 
There are several WG/RG pairs that may be considered for collaboration (NMRG/NMOP, MAPRG/IPPM, etc.) but I think that it would be better to scope the discussion to a specific area. I suggest to focus this time on NMRG/NMOP.
 
Thomas kindly accepted to drive a discussion on this topic in the IETF#126 OPSAREA Session with some proposals to bridge both communities. Thomas is also willing to collect feedback and coordinate between the NMRG/NMOP Chairs, in particular.
 
@NMOP/@NMRG Chairs: Would you be interested/willing to work with Thomas to consolidate some discussion material? Thoughts and suggestions are welcome.
 
Thank you.
 
Cheers,
Med
____________________________________________________________________________________________________________Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent doncpas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signalera l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci. This message and its attachments may contain confidential or privileged information that may be protected by law;they should not be distributed, used or copied without authorisation.If you have received this email in error, please notify the sender and delete this message and its attachments.As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.Thank you.
_______________________________________________
nmrg mailing list -- [e-mail]
To unsubscribe send an email to [e-mail]