At Mobile World Congress, Red Hat and NTT unveiled an initiative that enables real-time analytics at the edge based on large data sets. This is particularly useful in telecom.
The initiative uses technologies developed by the IOWN Global Forum, an organization focused on frameworks for telecom networks. The edge solution uses IOWN All-Photonics Network (APN) and data pipeline acceleration technology in IOWN Data-Centric Infrastructure (DCI). NTT’s accelerated data pipeline for AI applies Remote Direct Memory Access over APN for efficient collection and processing of large amounts of sensor data at the edge.
It then deploys the Red Hat OpenShift platform. This platform supports executing workloads in the data pipelines, whether geographically distributed or in a remote data centre.
Proof-of-concept
NTT and Red Hat have demonstrated that the solution can reduce power consumption and maintain low latency for real-time AI. The proof-of-concept (PoC) was evaluated with the sensor base in Japan’s Yokosuka City and a remote data centre in Musashino City. According to Red Hat, the latency required to collect sensor data for AI analysis was reduced by 60 per cent compared to conventional AI inference workloads, even when a large number of cameras were deployed.
NTT and Red Hat also collaborated with Fujitsu and Nvidia to enable the new data analytics technology at the edge. Fujitsu provided a PRIMERGY RX2540 rack server for the PoC, containing NVIDIA A100 Tensor Core GPUs and NVIDIA ConnectX-6 NICs. The PoC also uses Nvidia libraries for data pipeline acceleration.
23 hours ago
