CNF Taskforce ↔ ETSI collaboration (February 22, 2022)

Owned by Ranny Haiby
Last updated: Mar 28, 2022 by Xu Yang
2 min read

Proposed meeting logistics:

Feb-22-2022, 2PM UTC, 

Bridge info: https://zoom.us/j/97595070639?pwd=ajFTZGdlTmRVNjU5MSt1YVpycmlrdz09 (MODCOM bridge)

Contacts: ETSI NFV (Bruno), ONAP CNF TF (Catherine), ONAP MODCOM (Xu), ONAP-NFV Contact (Thinh)

Tentative Agenda (information sharing and no decision making):

- Brief update from ETSI-NFV project (5 minutes)

- Brief overview of CNF Direct project (5 minutes)

- Detail on ASD concept, IM/DM and packaging. (30 minutes)

- Q&A (20 minutes)

- AoB

Meeting materials





Meeting Record:

https://lf-onap.atlassian.net/wiki/download/attachments/16495201/onap-cnf-2022-02-22-ETSI-Meeting.mp4?api=v2

Meeting Minutes:

1) ETSI-NFV alignment work in ONAP (Byung-woo):

issue find:

  • vdu doesn't match well with vf-module concept in ONAP

  • purpose of S-VNFM is questioned

  • VNFD and Helm Charts have duplicate information

2) CNF direct path project (Lukasz):

3) ASD model proposal (Thinh and Marian):

  • ASD is an alternative approach compared with the current SOL001 VNFD model

  • one of the key intention is to avoid duplication with Helm

4) Q&A:

  • Bruno:

    • 1) in ETSI-NFV alignment, does the implementation imply no Helm Charts are included? A: no implementation yet

    • 2) how to support cluster selection among several compute resources? A: the helm charts can support them

  • Shitao:

  • Thinh:

    • clarify if the analysis is individual contribution? A: yes

  • Cristina:

    • in principle the 2 approaches are different, both from modeling and architecture perspectives. VNFD takes an abstract modeling approach for both VMs and containers, and covers subset of resources from native templates. Whereas the ASD only complements the native templates, and does not require the NFV-MANO architecture. Each has its Pros and cons, but I don't see how the 'merge the models' idea can apply here, these are conceptually different solutions.

  • Bruno:

    • merge of the models and merge of the orchestration methods are seperate topics

    • it would be good if we can have single model and orchestration methods in the industry

    • ask why Thinh is negative on having single model/orchestration method?

      • Marian: philosophy are different for ASD and VNFD

    • whether the info captured in ASD should be added into the VNFD

  • Hui:

    • agree with Bruno that we could have another session to further discuss the proposals, involving more ETSI delegates

    • there are many commonalities between the two models, not suggesting ONAP to copy the models while just changing the names

  • Kuno:

    • Because network information as Virtual Link is very important for telecom application, how to describe these network information by ASD and how/who can connect VNF to operator network which had several issue in past in ETSI-NFV? What is different concept between ASD and VNFD when consider these network aspects ?

    • What/How ASD can ensure backward comapatibility concept or mechanism?

  • Bruno:

    • suggest to use the gotomeeting bridge for another session

    • 4. Is the ASD really technology independent? there are several references to K8S in the IM

    • 5. You said ASD will support the CNF direct model. Can you elaborate on this?

    • 6. Will the changes made to SOL004 be proposed to ETSI? Will the new 5G non-MANO-artifact be registered?