June 28, 2017 Workflow review
Chengli Wang and Lingli Deng from China Mobile presented VoLTE use case. They went through the work flow for service design, instantiation, auto healing/auto scaling and termination. 31 Participants joined the meeting and discussion. The meeting recording can be accessed here: https://zoom.us/recording/play/zClGuWa2cFq-g3YEWC-sZ74V9iJaHAx2Iw7ePKuhh1q1ttjSSFXRyoJhYxpmk8bC
Meeting minutes:
At high level, there are edge and core data centers in the use case. EPC components split between edge and core data centers. Similarly IMS components also split between edge and core. A WAN network with underlay and overlay connects the two data center.
...
Need to talk to OOM and MSB teams to see if we need those modules in the use case.
July 13, 2017 Meeting with CLAMP, DCAE, Policy and Holmes teams
The meeting was to clarify workflow related questions with regard to CLAMP, DCAE, Policy and Holmes projects.
1. CLAMP team needed some clarification of DCAE and Holmes relationship on VoLTE design workflow. After some discussion teams agreed that we need to separate Holmes deployment scenarios into two options. First option is Holmes is deployed as an DCAE analytics; the 2nd option is Holmes is deployed as a standalone component in ONAP. The first option is what everyone agreed to support in R1, and the 2nd option needs more discussion in a separate meeting. The VoLTE workflow will be updated for the 1st option where DCAE should be labelled as DCAE collector, and Holmes should not be in the design time. Also, configuration policy is pushed by CLAMP to Policy engine first and then Policy engine forwards the configuration policy to DCAE. Closed loop template blueprint is sent by CLAMP to SDC and SDC forwards to DCAE.
2. StringMatch policy configuration shown in CLAMP VoLTE use case is not enough to support Holmes alarm correlation rules. CLAMP team needs to know what exactly needs to be configured for VoLTE use case. Teams agreed to start with concrete event example from a single VNF failure, and CMCC team will provide an example within 2 weeks. In R1, auto healing is required and auto scaling is an aspiration goal.
3. Team discussed DCAE deployment for multiple data center scenario. Ideally, some data collection and analytics function should be run close to data source. However, due to time constraint only one instance of DCAE is supported in R1.
4. In Auto-healing and auto-scaling diagram of VoLTE use case, modification is needed to show Holmes as part of DCAE as option one Holmes deployment scenario.
5. Explained that in VoLTE use case VF-C will collect VNF events from EMS and forward to DCAE in VES format. Also, MultiVIM component will report VIM events to DCAE directly in VES format.
A followup meeting will be set up for next Thursday.
July 13, 2017 WAN Overlay Network
...
5. MultiVIM project proposed to handle DC-GW through SDN local controller and expose APIs for ONAP to set up VXLAN overlay between data centers via EVPN.