A5G Networks and Napatech collaborate on high-performance, energy-efficient, Machine Learning-driven Multi-access Edge Computing optimizations, to be featured at MWC 2024

Press Release

Share This!

NASHUA, NH and COPENHAGEN, Denmark – Feb. 13, 2024 – A5G NetworksTM, Inc., a leader in distributed and autonomous 5G Core network software, and Napatech™ (OSLO: NAPA.OL), the leading provider of programmable Smart Network Interface Cards (SmartNICs) and Infrastructure Processing Units (IPUs) used in cloud, enterprise and telecom datacenter networks, today announced their collaboration on a high-performance, energy-efficient, Machine Learning-driven Multi-access Edge Computing platform for 5G networks, developed in partnership with Lenovo and powered by the latest AMD data center processors and adaptive computing products. While exhibiting at MWC 2024 in Barcelona from Feb. 26th through 29th, Lenovo (booth #3N30) and Napatech (#7B67) will showcase this solution in their booths.

Enterprises are increasingly deploying networking functions at the edge of the network instead of in the core, in order to save on backhaul costs, to minimize latency for real-time applications and to ensure that data remains on-site for regulatory compliance or privacy reasons. This trend, known as Multi-access Edge Computing (MEC), is especially applicable to industries such as manufacturing, transport, logistics, healthcare, energy and mining. The network edge is also where a massive amount of non-revenue-generating traffic is processed, such as video streams from TikTok and YouTube.

A critical function within MEC infrastructure is Local Breakout (LBO), which performs the offloading of mobile data traffic to a packet core running in the edge platform, enabling content and data to be processed at the edge so that the benefits of MEC can be realized. In order for this LBO function to be cost-effective, it’s important to maximize the number of devices and streams that can be supported in a small-footprint edge server, while minimizing the power consumption of server infrastructure deployed in remote locations. To address this need, A5G Networks and Napatech have collaborated to develop a MEC solution that leverages A5G Networks’ Machine Learning (ML) traffic detection functionality within its User Plane Function (UPF) for intelligently offloading specific network flows to Smart Network Interface Cards (SmartNICs) that implement the UPF data plane offload.

Available on a Lenovo ThinkEdge SE455 V3 server based on an AMD EPYC™ 8004 Series processor, the A5G-Napatech MEC platform delivers two key business benefits. First, offloading specific data traffic to a Napatech SmartNIC minimizes the utilization of general-purpose server CPU cores for this compute-intensive function, thereby maximizing the overall energy efficiency of the platform. Secondly, it provides significant savings in backhaul costs by creating separate slices for critical vs. non-critical network traffic, sending non-critical traffic to the public network and routing critical traffic to local Content Delivery Networks (CDNs).

“Our Autonomous 5G packet core is fully distributed and elastically scalable, using a cloud-native, microservices-based architecture to meet the growing demand of user plane traffic and processing,” said Kartik Raval, Co-Founder and VP of Technology at A5G Networks. “By deploying our ML-based traffic steering within the MEC solution that we have developed with Napatech, we’re able to improve energy efficiency by 8x and reduce server CAPEX by 86% in a typical edge use case, while also reducing backhaul bandwidth requirements by 67%.”

“Napatech’s 5G UPF offload sustains 100Gbps of full duplex traffic on a single NT200 SmartNIC based on an AMD Virtex™ UltraScale+™ VU9P FPGA, while processing more than 140 million concurrent flows with a flow learning rate greater than 1.5 million flows per second, thereby achieving a total throughput of up to 85 million packets per second and ensuring full wire speed operation for typical packet sizes,” said Charlie Ashton, Senior Director of Business Development, Napatech. “By completely offloading the UPF data plane from the server CPU, this solution enables A5G’s packet core to deliver industry-leading energy efficiency and cost-performance for MEC applications.”

“We’re pleased that A5G and Napatech selected the Lenovo ThinkEdge SE455 V3 Edge-optimized server to host their Machine Learning-driven, Multi-access Edge Compute platform,” said Dominique Vanhamme, General Manager, Lenovo CommSP. “Lenovo’s ThinkEdge SE455 V3 leading performance and sustainability means Communication Service Providers can benefit from accelerated packet processing and increased workload consolidation with 2x more storage and 30% less power consumption for the Napatech NT200 SmartNIC than any other edge server.”

“The performance and feature set of the AMD Virtex UltraScale+ VU9P FPGA enables Napatech to deliver impressive UPF offload throughput and functionality for evolving 5G infrastructure,” said Kumaran Siva, Corporate Vice President, Strategic Market Development, AMD. “When an AMD EPYC CPU-powered server is used in conjunction with our FPGAs, more CPU cores can be freed to run business-critical applications while maximizing energy efficiency. We are delighted that Lenovo and Napatech will be featuring this solution at MWC 2024.”

For more information on the energy savings delivered in this use case, see the white paper from Heavy Reading “Power & Energy Efficiency Strategies for 5G Networks”.

About A5G Networks
A5G Networks Inc. is a leader and innovator in autonomous mobile core software. The company is headquartered in Nashua NH, USA with offices in Pune MH, India. A5G Networks is pioneering secure and scalable 4G, 5G and Wi-Fi software to enable distributed and autonomous network of networks. For more information, visit www.a5gnet.com.

About Napatech
Napatech is the leading supplier of SmartNIC and IPU solutions used in cloud, enterprise, and telecom datacenters. Through commercial-grade software suites integrated with high-performance hardware, Napatech accelerates network infrastructure, security and storage workloads to deliver best-in-class system-level performance while maximizing the availability of server compute resources for applications and services. Additional information is available at www.napatech.com.

No Forward-Looking Statements
This press release may contain forward-looking statements which are only predictions and may differ materially from actual future events or results due to a variety of factors, including but not limited to, business conditions, trends in the industry and markets, global economic and geopolitical conditions, macro-economic factors, and other risks and uncertainties set forth in Napatech’s reports. The matter discussed in this release is based on current expectations and maybe subject to change. Napatech will not necessarily update this information. For details, visit us at www.napatech.com.

AMD, the AMD logo, EPYC, Virtex, Ultrascale+ and combinations thereof are trademarks of Advanced Micro Devices, Inc.