NMRG 50th meeting IETF 103, Bangkok, Thailand RG Chair: Laurent Ciavaglia Available During Session: Slides: https://datatracker.ietf.org/meeting/103/session/nmrg Meetecho: http://www.meetecho.com/ietf103/nmrg/ Available Post Session: Audio recording: https://www.ietf.org/audio/ietf103/ Video recording: https://www.youtube.com/user/ietf/playlists *********** Agenda: 1. Introduction, NMRG Chair 11:20, 5 min. 2. Progress on current research agenda 2a. Summary of Intent Based Networking topic and workshop at IETF 102 11:25, 10 min. 2b. Summary of Artificial Intelligence topic and sessions at IETF 102 11:35, 10 min. 2c. Summary of Autonomic Networking 3.0 topic 11:45, 10 min. 3. Discussion on the future of NMRG Open mic discussion Wrap-up and information of way forward 11:55, 25 min. *********** Summary: The meeting covered two aspects: 1) progress of ongoing research items: intent-based networking, autonomic networking, and use of AI techniques for network management – and 2) evolution of the RG (“re-chartering”). On the Intent based network work: comments from the room highlighted the work is important and could deliver good results, useful for the IETF and the industry at large. However, some comments pointed to the document on “terminology”, concepts and definitions stating as starting too late because solutions in the industry are already being developed and deployed, and the technology is discussed in the engineering community. Another comment highlighted the need to perform intent validation (across the “layers”, “models” and across “domains”), correlating telemetry with intent. On autonomic networking work: comment expressed some work needed to define how the ANIMA/ACP is applicable and working in case of multiple distributed autonomous entities, yet the overall system is still seen as a unified one. Comment to have a look at the work from the DTN WG which has experience with autonomy for communication. Proposal to define the self-CHOP, use cases and technologies applicable. A previous, now expired draft was going into that direction. Comment also mentioned other groups active such as ETSI ENI, collaboration encouraged. On use of AI techniques for network management work: due to issue with the remote presenter, the chair read the slides. Comments received were positive and supportive of the work in the area. Also pointed the RG should deliver something now. Comment pointed the central role of data in the AI/ML techniques and need to have work on the use of data in this topic. On the future of NMRG, the chair gave an introduction about some important questions (collective goal, type of output, beneficiaries…) , working principles (pluri-disciplinary and cross-fertilization) and introduced some topics. The discussion highlighted different viewpoints on the role and timeframe for the work of a RG (short/longer term research, awareness to IETF, address current problems/look into new stuff…). Beyond good topics, good resources are needed to bring work forward and to completion e.g. work on autonomic networking started in NMRG that finally led to the creation of the ANIMA WG. Linkage between WG and RG should be fostered. Another comment highlighted the need to both progress on the ongoing items and bring new ideas/topics. The chair closed the discussion by stressing that a transformation of the RG is ongoing, we should discuss/work on it but most importantly conclude the transformation and deliver. *********** Minutes: Thanks to Dean for the minutes, and Jéferson for the Meetecho scribe DB: Dean Bogdanovic (Volta Networks) JT: Jeff Tantsura (Apstra) JN: Jeferson Nobre BL: Bing Liu (Huawei) PMJ: Pedro Martinez-Julia (National Institute of Information and Communication Technology Japan) RT: Rick Taylor (Airbus) WL: Will Liu (Huawei) PL: Paolo Lucente LC: Laurent Ciavaglia Agenda bashing discussing research agenda presented in the ieee article 6-7 active documents please look at the documents to discuss the future of the NMRG most meetings co-located with IETF, one possible outside of IETF, IM 19, Washington 19 reach out to IEEE and similar communities poll to find out about the interest in other co-located meetings, AI, operator meetings (NANOG, RIPE) Summary of the intent-based-networking-00 presenter Jefferson Nobre DB: taxonomy of the terms. little bit too late, actively discussed in the engineering communities. JT: You need cross-validation between different models and there is lot of work between different models (service, device) being done. JN: We need clear definition of intent for use in the IETF and IRTF. There is division between different types of intents which have to be defined Summary of the Autonomic Networking presenter: Jeferson Noble Dean: thing to do urgently. new perspectives into network management requirements e.g. the ACP: highly distributed if each entities act autonomously, still seen as a unified system. this is an intermediate step needed. Rick: deep space communication large experience on autonomy. DTN . If you want to learn more about, come tomorrow to DTN tomorrow morning Bing Liu: propose new work. mention self-CHOP. propose taxonomy on self-CHOP. Look at different scenarios where self-CHOP could be used and how other technologies can help. JN: There is a draft from Costas from long time ago was about the topics BL mentioned. We can discuss about this Will Liu Huawei: Supports the work on this topic. New consortium was started to address this problems and is interesting to cooperate with NMRG Extending Intelligent Network Management Beyond SW and Logical Boundaries presenter: Pedro Martinez-Julia problem with the audio connection and the presentation was cut Laurent Ciavaglia continues presenting the slides due to technical difficulties. Apologizes for not being able to provide deeper view of the document and answer the question Rick Taylor (Airbus): Good presentation, good points, we have to answer some of the questions from the document. We have to go and do it somehow. Here or in another SDOs WL: All the ML/AI are based on data. There should be work done to find out other ways Talk about possible new co-chair soon What is the future of the NMRG? presenter Laurent Ciavaglia Some ideas and topics –Managing at ultra-low and high scales and speed –Role and place of human (in the (management) loop), human-network interactions, instrumentation, visualization, trust/oversight –Post-NETCONF/YANG era –Site/Network Reliability Engineering field (SRE/NRE) We have lot of freedom how we work Dean: timeframe RT: No timeframe should be defined, looking what is happened. Linkage bw IETF/IRTF. Awareness of what's happening, guide WGs DB: mid-term things get lost and surprise can happen to the IETF RT: a good balance should be JT: it is cool to jump into new things, there are some stuff to be solved and directions is needed. Help the IETF LC: this is one of the questions should be addressed DB: as individual and NOMCOM 2018. ask candidates about linkage IRTF/IETF. inputs to IETF very much important. awareness PL: I'm supporting DB comments as from my work running into issues as there is no standardization in certain areas RT: agree with DB. Nice summary. IRTF should foster research and raise awareness LC: Topics are very important. Resources are also very important. Different people from different backgrounds providing info. We need people to contribute, not just open and present problems. RT: IETG WG chair hat on. What IRTG RG are working? Ask IETF to chip in to the work LC: there are examples of well collaboration, good example is ANIMA WL: one is about continuity, like intent. The other one is about bringing new work to the RG LC: it will take time to transform this RG. We have to conclude what is our plan and then implement it. I'm trying to steer this transformation. Your input is important. Please contribute to this process **********