You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 67 Next »

Meeting Minutes 

Jan 4th, 2022 at 2pm UTC 

Jan 18th, 2022 at 2pm UTC

Jan 25th, 2022 at 2pm UTC

Feb 1st, 2022 at 2pm UTC

  • Chaker Al-Hakim Sharing suggestions for the CNF landing Page (E2E flows, HL diagrams)
  • Do we want to move the comments from the landing page to 2020 MoM etc?

Feb 8th, 2022 at 2pm UTC

  • Update on ASD
    • Result from [onap-modelingsub] Email Poll on ASD Information Model andPackaging Format Marian Darula , @Thinh Nguyenphu
      • Huawei is objecting  the ASD concept on grounds that it is not aligned with ETSI-NFV
      • During the poll there were no merit based objections. The only objection was the procedural one above.
      • ASD is seeing support in the O-RAN community. Thinh Nguyenphu: Dropping ASD in ONAP might hurt it and put it behind other projects.
      • Marian Darula - There is a need to escalate the issue and reach a decision.
      • What could be the alternative? -
        • Fernando Oliveira : The ETSI proposal is capable and satisfactory, providing the necessary functionality without significant complexity.
        • Byung-Woo Jun : There was an attempt to align with ETSI. There were some challenges converting policies to VF module. The conclusion was this attempt  did not work well.
        • Marian Darula : There is a challenge synchronizing content between Helm chart and VNFD. That was the main concern by Ericsson that drove to the development of the ASD.
        • Thinh Nguyenphu : Another  motivation for ASD was parametrization, and the desire to  use cloud-native approach to that.
      • Can there (and should there) be one solution for packaging? Marian Darula - It was clear from day-1 of the ASD effort that there might be more than one approach. There was never an attempt to make ASD the only winning solution.
      • Marian Darula , Thinh Nguyenphu : If ONAP does not adopt ASD, the development effort might move to another community, leaving ONAP "behind".
      • Huawei is not objecting continuing the ASD PoC. 
      • Kenny Paul : The modeling subcommittee can make recommendations to the TSC, but ultimately it is a TSC decision.
      • POC definition Approved PoC policy since the Dublin release. A subcommittee cannot block a PoC,. All that is required is that the PoC needs to be reviewed with the committee.
      • Ranny Haiby : Any request to the TSC should be clear about keeping both approaches alive, and let them be developed in parallel, as alternatives.
      • Next  step: Bring this to the TSC to decide the future of the ASD PoC.
      • Timo Perala notes that REQ-993 - Getting issue details... STATUS is listed in Jakarta Release Requirements
        • note: this requirement covers ASD IM/DM. The ASD onboarding, distribution and LCM orchestration plans to be handled as a PoC during Jakarta. - Byung
    • Network Service modeling/ASD - Latest Updates - Byung-Woo Jun , Fernando Oliveira 

Feb 15th, 2022 at 2pm UTC

  • Clarify the benefits of ASD  over the ETSI-NFV SOL001
    • There was a  meeting on feb-14th on the modeling subcommittee  - no consensus reached - Modeling Subcommittee 2022-02-14
    • Decision was to escalate the decision to the TSC. The request from the TSC is to enforce adherence to the modeling subcommittee procedure. Andy Mayer - CLEAN state is a result of a subcommittee decision, based on rough consensus. The poll is just an input to the rough consensus. Xu Yang - "rough consensus" means there are no un-addressed objections. Marian Darula - the objection cannot technically be resolved.
    • Xu Yang -  Based on input from ETSI experts, it may be possible to unify the models.
    • Zu Qiang (Ericsson) - Why is there an attempt to create a single model/package?  Xu Yang - There is no such limitation to have one model. The ASD model is referencing ETSI models.  Therefore it seems that it should be possible to find a unified model. For example, the ASD may become part of the ETSI-NFV NSD. Thinh Nguyenphu -  ASD is not "built around" ETSI models. It may reference ETSI, just like it may reference any other SDO. It may be a good idea to aspire for more alignment, but it  is inappropriate to block progress on the ASD initiative. Requests to follow the letter of the governing document (Approved ONAP Model Governance), and decide based on the poll.
    • Byung-Woo Jun - The ASD PoC is NOT using elements from ETSI (e.g., SOL005, SOL003, NFVO, VNFM), so it cannot reconcile, as it will require major changes in the ETSI specifications. The "ONAP Native" orchestration approach is planning to adopt ASD,  and is also not using ETSI specs. Why not use the same approach for the ASD and allow it to exist as an alternative?  Xu Yang - The ONAP Native approach was approved because there was no reference to ETSI and it seemed like a stand alone approach.
    • Xu Yang - Expectation from the TSC - Decide whether the current ASD proposal is technically mature enough, or does it need more work for reconciliation with ETSI?
    • Timo Perala -  Is there an estimated expected time line for the potential ETSI NFV - ASD aligned solution? For all I know it may easily take until 2023 to get there.
    • Catherine Lefevre - What if ETSI endorses the ASD? Byung-Woo Jun - It  will require major changes in the ETSI MANO  architecture, may take time as stated above.
    • Xu Yang The ASD IM does make references to ETSI.  There are references to Virtual Links for example.
    • Zu Qiang (Ericsson) - ASD Information model does not make a reference to ETSI. It may use identical attribute  names, but that can be changed to avoid confusion.
    • Thinh Nguyenphu -  TF (Catherine), ONAP MODCOM (Xu), ONAP-NFV Contact (Thinh) Bridge info: TBD Document depository: TBD (my recommendation is CNF TF wiki page) Tentative Agenda (information sharing and no decision making):
  • Prepare for Joint workshop with ETSI-NFV regarding ASD

Feb 22nd, 2022 at 2pm UTC

Mar 1st, 2022 at 2pm UTC

  • Prepare for LFN CNF workshop on Mar 13/14 - https://wiki.lfnetworking.org/display/LN/2022+LFN+DTF+Workshop+Topics
    • Catherine Lefevre will work with EMCO (Bob Monkman) to prepare an integration proposal. Lukasz RajewskiRanny Haiby  will take part in the discussion

    • The goal is to have a clear roles and responsibilities definition for cases like the 5G super blueprint

  • ASD Model approval - Can we come up with a compromise proposal that will put everyone's mind at ease?
    • Focus on the technical aspects, not the procedural and vision aspects .
    • Keep the momentum going and avoid future re-work.
    • Path to move forward:
      • Work on 'definition" problem raised by Fernando Oliveira , Zu Qiang (Ericsson) , Marian Darula  + Thinh Nguyenphu
      • Position ASD Model as an alternative to current ETSI Standard proposal to move forward with "Cloud Native Architecture"
        • ASD implemented as plug & play (you choose onboarding ASD / ETSI / others)
        • Known Impacts: SO e.g. CNF O (particular Camunda) + SDC /AAI (to be determined by POC work) + particular VES Collector - Byung-Woo Jun to review this part 
      • Work with the Modeling Team Catherine Lefevre 
      • Continue to share to ETSI what we do Thin, Byung-Woo Jun  - increase visibility

Mar 8th, 2022 at 2pm UTC

  • Follow up on ASD Model Action Items
    • Work on 'definition" problem raised by Fernando Oliveira , Zu Qiang (Ericsson) , Marian Darula  + Thinh Nguyenphu.
      • Presentation by Marian Darula - <please upload deck here>
      • Reviewed the difference in approach between ASD and ETSI - ASD adds a "slim" descriptor that only fills the gaps of the Helm charts. The ETSI approach creates a full SOL001 IFA11 descriptors with the full set of attributes. The ETSI approach may lead to duplication of information and two separate entities trying to control the lifecycle of the CNF (orchestrator and K8S)
      • The ASD approach is using the IM of K8S/Helm
      • Fernando Oliveira - Contention point is about the variation between IM and implementation. The old ECOMP model or ETSI model is satisfactory,  there is no need for a new IM.
      • Zu Qiang (Ericsson) - Explains the xNF modeling with ASD
      • shitao li - There is no real duplication with the ETSI approach - Many of the attributes are  optional. The comparison of LCM approaches may not be accurate
      • Thinh Nguyenphu - In reality, it is not possible to use IFA11 to model a CNF.  Ericsson tried that in the last two releases and proved it cannot  be done. ONAP does not  have a general CNF IM. The ASD goal is to simplify the onboarding of CNFs, removing the need for VNFM
      • Thinh Nguyenphu - Proposes to respond to Fernando Oliveira 's comments
      • Chuyi Guo - It sounds like there is a need to continue the technical aspects in the modeling subcommittee. Proposes to use the upcoming meeting for this technical discussion.
      • Thinh Nguyenphu , Marian Darula - What is the point of continued discussion?
      • Catherine Lefevre - We need to move forward with implementation. Let's treat the ASD IM as v1, and work towards a v2 that will address the concerns of the modeling subcommittee.
      • AI  - Marian Darula - upload the existing ASD model as a v1
      • AI  - Catherine Lefevre - Will follow up in the TSC.
  • Presentation ONAP/EMCO Alignment prior LFN DDF
  • No meeting on Mar 15th - replaced by LFN CNF workshop

Mar 22nd, 2022 at 2pm UTC

  • Feedback from the LFN DTF Event:
    • CNFs
    • EMCO/ONAP Alignment
      • There are clearly two profiles. Not clear how to make progress on both
      • Issue with profile #2 - How to use the policy functionality of ONAP, without the inventory components.
      • Seshu Kumar Mudiganti - We should go to the next level of details on the integration points between ONAP and EMCO - We need a simple use case. 
      • What use case should we start with? We should find a use case that demonstrates the value of the integrated solution. It should showcases functionality that is not available with ONAP-only or EMCO-only.
      • Seshu Kumar Mudiganti - Who is responsible to turn the wishlists to implementation? Who will take responsibility?
      • Lukasz Rajewski - What is the right motivation for profile #2? Is it to make the solution lightweight? Is it to address edge use cases?
      • Vishal Sharma  - Initial use case UDC on a single cluster or multiple clusters. Might be on-prem (OpenShift) cloud or public cloud. Lukasz Rajewski - Deployment on public cloud might require EMCO, as it is not currently supported by ONAP. Seshu Kumar Mudiganti - No guarantee upfront that the UDC will work out of the box. Experience shows it might require some tweaking. Seshu Kumar Mudiganti - Suggests to start with ONAP-only, then identify gaps where EMCO may be useful.
      • Vishal Sharma - Slicing scenarios might require multiple clouds, which calls for EMCO. Lukasz Rajewski If 3GPP slice API are required, they are not provided by EMCO. ONAP does provide this API.
      • Vishal Sharma - Mentioned the AMCOP solution from Aarna Networks, that uses EMCO API for 5G Core deployment, and ONAP CDS for Day-1/2 configuration.
      • Vishal Sharma - How can ASD be used today? Especially when ETSI is still working on specifications? Marian Darula - There is documentation and on-going PoC. Byung-Woo Jun is the main contact person for more information. On-boarding and deployment are expected to be completed in the Jakarta time frame. PoC wiki space - ASD-Based CNF Orchestration PoC
      • Vishal Sharma - Will provide future updates on his progress using the CNF Taskforce mailing list, wiki, or meeting.
    • Next Steps
  • Decision on new meeting time
    • Starting next week the meeting will be at 13:00 UTC
  • Are we good with ASD Information model? What are the next steps?
    • No immediate blockers.
    • @thinh - There is a planned workshop with ETSI today. At the moment there is no plan for further workshops


Mar 29th, 2022 at 1pm UTC (← New time!)

  • ASD PoC Status update - Byung-Woo Jun 
    • SO and SDC teams from Ericsson are working on the PoC
    • SDC team is working on the resource VF. working on TOSCA parser. Map ASD into VF, Wrap in CSAR and distribute.
    • SO team working on LCM swagger definition for northbound API. Waiting for creation of PoC branch. Focusing on calls from BPMN to CNFM.
    • Lukasz Rajewski - Will the ASD LCM API be integrated with the rest of the LCM API? 
    • The PoC handles onboarding and instantiation with variables.
    • Slide deck - Byung-Woo Jun , ASD-PoC-Status-20220329.pdf
    • Vishal Sharma - ETSI is expected to come up with SOL003 specifications, is there work planned to support it? Byung-Woo Jun , Fernando Oliveira - There is definition work ongoing for the ETSI approach, but actual development has not yet started.
    • Target for PoC - Initially was Jakarta, but now it is clear some functionality will slide into Kohn

Apr 5th, 2022 at 1pm UTC 

  • Updates from ETSI workshop - @Thinh

Action Item(s) (In Progress)

  • (CNF Task Force): What do we need to ask to CNF Vendors to be onboarded on the ONAP Platform? These reqs could be shared with Anuket Assurance for the CNF badging

Action Item(s) (Closed in 2022)


  • No labels