In R2, we plan to support the following use cases for Integration/E2E testing (use cases moved from OOF Beijing M1 Release Planning):
- Use Case: Residential Broadband vCPE (approved)
- Hardware Platform Enablement In ONAP
Use Case: VNF scale out (and potentially service scale out)Use Case: VoLTE (approved)5G family of use cases- Broader CMSO use case
Note:
...
- SDNC was moved out the scope for R2 for the instance capacity checks. However, the plugin that exercises this capacity check with SDNC exists in the repo, but is not exercised.
- Similarly, Reservations (with MultiCloud and SDNC) were moved out of scope from R2, and into R3 due to lack of resources.
Proposed Test plan
As discussed, Here's list of things we need to discuss and address:
* placementDemands should include the vgMuxInfra and vG
* BRG should be removed from the placementDemands since it is a PNF and homing doesn't really make sense given its location is already fixed.
* the licenceDemand section should be removed (potentially) since we aren't really doing any license optimization in R2.
Also, the testing would involve two incremental steps:
Step 1. Test the basic Homing workflow for vCPE with the air-distance and colocation constraints (https://lf-onap.atlassian.net/wiki/display/DW/HAS+%28R2%29+Beijing+Release+planning).
Step 2. Add the HPA and MultiCloud capacity check policies, and test with the functional test cases.
The first step would ease us into the workflow replacing the SNIRO emulator, while step 2 would test all the functional features we support in R2.
Thanks,
Shankar.
Gaps
- Customer location: Ideally, the PNF (BRG) location should be used, but based on Kang Xi's recommendation for R2 (in R1, customer location was hardcoded in homing emulator), we have this come as a part of the order parameters from SO. Addressed - this would be obtained through SO in R2.
- vGMUX Service Instance availability in A&AI. Our implementation uses v11 API, but how is this information populated in AAI ? [ addressed this by an alternate route to get the complex, since pservers are not being created in R2 ]
- Testing HPA: vG is the only possible candidate, since that is the VNF that is spun up in this workflow (vgMuxInfra already exists, and we pick an existing instance as a part of the homing solution). Are there other relevant VNFs ?
- Homing policies need to be manually created for the vCPE service in R2. Ideally, homing policies should be created in the policy framework from the SDC models, and consumed by OOF.