Red Hat, Inc., and SoftBank Corp. (“SoftBank”) today announced the implementation of AI-RAN to optimise power consumption and networking performance using Red Hat OpenShift, a leading hybrid cloud application platform powered by Kubernetes.
With this collaboration, Red Hat and SoftBank are addressing many of the long-standing RAN implementation challenges that service providers often face, such as balancing user demands with energy costs, resource availability and managing deterministic and distributed workloads.
By bringing together AI and RAN on a common platform, Red Hat OpenShift, service providers can dynamically adjust network parameters to meet changing demand and streamline network operations for higher agility.
Enhanced Network Optimization
SoftBank is working with Red Hat to develop AITRAS – an integrated AI and RAN solution built on Red Hat OpenShift. AITRAS provides an enhanced network orchestration and optimization solution that can support virtualised RAN and AI-enabled applications alike, enabling service providers to operate diverse applications with greater consistency and flexibility.
Additionally, SoftBank is working with Red Hat to use community-driven technologies, like Kepler, an open source project founded by Red Hat, to help service providers reduce energy costs by more accurately capturing and act upon power-use metrics from applications.
The power monitoring capabilities of Red Hat OpenShift are derived from Kepler, which exposes key metrics at the cluster level to probe key performance counters and other system statistics.
These metrics can then be fed into SoftBank’s AITRAS orchestrator to help even out power consumption across disparate sites and optimize energy. Using Red Hat OpenShift and its power monitoring capabilities.
According to Ryuji Wakikawa, vice president, Head of the Research Institute of Advanced Technology, SoftBank Corp “Electricity and telecommunications services continue to grow as critical infrastructure that supports society. By monitoring and predicting power consumption, ‘AITRAS’ optimizes equipment from an energy efficiency perspective while reducing risks through distributed deployment”.
Advantages of AITRAS:
- Optimise the placement of compute- and GPU-intensive workloads according to power consumption metrics to maximise energy usage across disparate environments, while still maintaining enhanced performance.
- Enable lightweight measuring using Red Hat OpenShift and Kepler, via extended Berkeley Packet Filter (eBPF) for kernel-level data collection, to help reduce overhead and improve energy efficiency.
- Achieve more precise GPU energy calculations for both Linux processes and Kubernetes pods using enhanced observability capabilities with Red Hat OpenShift and Kepler, enabling service providers to pinpoint granularities across multi-instance GPU (MIG) and various GPU devices.
READ MORE ABOUT AI DRIVEN CLOUD SOLUTIONS: READ HERE
