Minutes

OVP Submissions

• ZTE Lab submission:

  • Still need technical review of the results.

  • @Lincoln Lavoie will assign specific reviewers to handle this review this week.

• Review of submissions is taking a long time, and seems to require directed calls (i.e. ask specific people to review)

  • Program was founded on the idea of community review, similar to code review, etc.

    • OVP Reviewers

  • We can start assigning reviewers, but this would seem to run counter to the program intention (i.e. community review process).

OVP 2019.12 Release

• OVP 2019.12 release checklist 

  • Consider pushing back the OVP2019.12 release until after the January plugfest / DDF.

    • If we release after the Plugfest / DDF, it's harder to message the outcome of the plugfest can be certified VNFs.

    • Group agreed to keep the December release target

    • We need to reach out to @Dan Xuand the Dovetail team to get assignments and tracking for the release documentation.

Beyond OVP 2019.12 Release

• Is orchestration always required to do VNF validation testing?

  • Would this split VNF validation into two "camps" for with and without orchestration (i.e. ONAP)

• Performance Testing

  • The "hard question," how to test the function of a VNF without prior knowledge of the VNF and "what it does."

  • Could the test environment (i.e. the test setup, packet generator, etc.), where VNF participants would need to submit a standardized "traffic template" along with their VNF for testing.

    • The template will become critical, as this needs to be consumable by the tools and used by others to "check" or "validate" the results.

    • Could this leverage the BBF work to develop Application Layer Testing, which will create data models (information models) for this type of traffic definitions.

• @Rabi Abdel is going to bring in the updated slides that show where testing for be extended in the OPNFV and ONAP cases, taking into account CNTT input, etc.

  • Attachment: 

  • This will be discussed next week, so please review and bring questions and feedback.

• How is testing invoked in a "user" friendly way?

  • How can this be improved?

  • Avoid purely manual processes?

  • Can we (should we) define guidelines or best practices for how testing is run? For example, could a norm be to use a docker container, that is pulled from a reliable repo (only an idea as an example, not a proposal)?

Any other business

• Meeting Time:

  • @Lincoln Lavoie and @Rabi Abdel  to discuss with Bin moving either the CVC call or the OPFNV tech discuss call to avoid the current overlap caused by the time-zone change.