Rakuten Symphony unveils revolutionary network monitoring solution powered by eBPF Technology

BARCELONA – In a ground-breaking development, Rakuten Symphony is proud to announce the successful implementation and testing of its innovative eBPF technology within the fully virtualized radio access network of Rakuten Mobile. This is a major leap in network management and performance monitoring solutions, addressing critical pain points faced by the industry.

As part of the Rakuten Symphony technology stack, eBPF has been seamlessly integrated into its Observability Framework, enhancing performance monitoring, event tracking and anomaly detection across all network domains. This integration further drives near-real-time lifecycle automation and orchestration actions, end to end.

"The application of eBPF has revolutionized networking, enabling enhanced performance monitoring across diverse network domains, active testing on multiple devices, and AI/ML-driven automation for anomaly detection and network bottleneck prediction, said David Soldani, SVP Innovation and Advanced Research, Rakuten Mobile. “Our solution transforms traditional observability frameworks on modern Telco networks, offering a more agile, efficient and intelligent approach."

Traditional transport-network-layer monitoring systems have long been hampered by high costs, limited interoperability and inadequate protocol support, among other challenges. Rakuten Symphony's innovative application of eBPF technology marks a significant departure from these constraints, heralding a new era of agile, efficient and intelligent network infrastructure management.

Rakuten Symphony's eBPF for transport network layer allows for performance monitoring across all network domains, introduces active testing capabilities on a diverse range of devices and harnesses the power of AI/ML to automate anomaly detection and bottleneck prediction, among other relevant use cases. This holistic approach not only streamlines network management but also ensures a more reliable assessment of network performance, addressing issues such as latency, jitter, packet loss and quality of service(QoS) head-on.

The implementation of eBPF for performance monitoring at the transport network layer has been validated through practical tests conducted on Cisco routers. The eBPF solution successfully measures latency fluctuations, offering instantaneous and valuable insights into the network's robustness during periods of stress.

Additionally, eBPF agents were deployed across the radio access network infrastructure, including servers hosting virtual distributed units (vDUs) and virtual central units (vCUs) within the radio access network (RAN). These deployments further validate the effectiveness of eBPF across various network segments while imposing minimal additional load on the CPU.

Rakuten Symphony's successful application of eBPF technology within the Rakuten Mobile network infrastructure demonstrates our commitment to leading the charge towards a smarter, more connected future. Rakuten Symphony eBPF is poised to transform the landscape of network monitoring by offering a versatile, cost-effective and user-friendly alternative to traditional systems.

Furthermore, we are developing additional solutions including:

• eBPF for performance tracing of any aspect of a system, achieving granular visibility into network activity, resource utilization and traffic monitoring. This includes AI/ML automated prediction of network, database (NFS) and cloud-native performance.
• eBPF for monitoring and optimizing network resource utilization and energy consumption, along with AI/ML automated prediction of power dissipation and energy consumption of any CNF (or microservice, interface) and capacity forecast in Kubernetes clusters.
• eBPF for real-time threat detection and response, as well as eBPF for runtime security to enhance security tooling, using ML Agent & eBPF as data sources & inference. This includes AI/ML automated anomaly detection from log files, and real-time threat detection and response to unauthorized access.