Network Traffic Analytics through Accelerator and SmartNIC at Edge
Network traffic analytics is critical to data center management and operation efficiency. The regular solution with cloud-based analytics has higher latency and delay for action feedback. This paper p …
| Talk Title | Network Traffic Analytics through Accelerator and SmartNIC at Edge |
| Speakers | Mrittika Ganguli (Ganguli, Intel Corporation), Yuming Ma (Cloud Solution Architect, Intel Corporation) |
| Conference | Open Networking Summit Europe |
| Conf Tag | |
| Location | Antwerp, Belgium |
| Date | Sep 23-25, 2019 |
| URL | Talk Page |
| Slides | Talk Slides |
| Video | |
Network traffic analytics is critical to data center management and operation efficiency. The regular solution with cloud-based analytics has higher latency and delay for action feedback. This paper proposes an edge-based analytics model with close-loop action feedback to achieve real-time predictive network analytics. Under this solution, the network traffic telemetry data is collected and aggregated by the network devices (e.g. SmartNIC). The AI model is trained at cloud and deployed to the SmartNIC. The predictive analytics and AI inference is managed at the edge device of SmartNIC to achieve real-time close-loop management and mitigation.
