Overview
From the perspective of D release and beyond in CCVPN use case, some users have the requirements of specific value-add services, such as intelligent surveillance, in which centralized monitoring portal will be deployed at the edge DC near the enterprise HQ, and it will have AI applications for collecting and analyzing voice or video monitoring data for abnormally recognition, explosion detection and so on. The AI applications will be deployed in a distributed way to the edge DCs that are near the specific site(s) under surveillance.
Since CCVPN has the requirement in large amount of voice or video monitoring data to collect and analyze, it brings the demand in accelerating this process using some specific acceleration technologies in AI field, such as GPU, FPGA ,etc. And it requires a special automated analytic mechanism for past data in a specific observation cycles in existing ONAP.
AM can deal with the problems CCVPN encounters.
BP in Dublin and beyond
From users’ perspective, we have drawn a BP of AM in D release and beyond. The detailed workflow is shown in ‘Workflow & Impacts’ part :
Discovery and registration of acceleration resources.
Instantiation and assignment of acceleration resources .(Reuse existing process in HPA)
The termination of the use of acceleration resources and release process.
Adjustment of suitable acceleration resources according to specific case with the data AM monitored and analyzed. – Whether to put it in D release or beyond depends on different target of different cases, seen as stretch goal.
Participating Companies
China Mobile, Lenovo, Huawei, Wind River, Intel
Workflow & Impacts
It is envisioned that AM might have impacts on the following projects: VFC, Policy, MCloud, AAI, VNF SDK, OOF and DCAE.
The detailed workflow as follows:
- Discovery and registration of acceleration resources
- Instantiation and assignment of acceleration resources
- The termination of the use of acceleration resources and release process
- Adjustment of acceleration resources
