Introduction

Businesses today face significant challenges in enhancing customer experiences while having to minimize their total cost of ownership (TCO). Such challenges bring about the need for innovative solutions that not only ensure security but also improve service efficiency as traditional IT infrastructures often struggle to keep up with the demands of modern, customer-facing applications. These   requirements are driving a shift towards edge AI, where processing occurs closer to the data source, in real-time, with reduced latency and enhanced privacy.

Challenges

Infrastructure Limitations:

Businesses encounter technical difficulties when trying to deploy edge AI applications around their existing infrastructure and processes. Existing IT setups often lack the computing capability to handle advanced AI applications efficiently due to outdated hardware, leading to increased latency and reduced effectiveness of customer-facing technologies.

Total Cost of Operation:

The separation of 5G telecommunications from AI computing infrastructures proves inefficient. Integrating AI workloads with 5G networks using an unified high-performance computing platform enhances technical capabilities and reduces costs relating to equipment, power and space.

Customer Experience:

5G service providers need to deliver personalized, time-sensitive services that fulfill customer expectations, therefore, as already mentioned, driving a shift towards edge AI, where data processing takes place closer to the source in real-time with reduced latency and enhanced privacy.

Solution

Consolidating AI at the edge with 5G using an edge AI platform turbocharges digital transformation and enhances value creation. With industry-leading expertise in delivering enterprise AI on GPU-accelerated platforms, Lanner is well-positioned for this new computing paradigm. These platforms offer enterprises, telecommunications operators and cloud service providers (CSPs) the opportunity to seamlessly integrate 5G with edge AI ecosystem, or to convert the 5G gNB to an edge data center for AI workloads.

The ECA-6040 is a powerful 2U short-chassis edge AI server designed for real-time inferencing and training at the 5G edge. It offers a GPU-accelerated, 5G vRAN-enabled computing architecture delivered as a software-defined platform consolidated with other AI workloads. This platform supports up to 1,024GB DDR5 system memory and leverages the 4th Gen Intel Xeon Scalable Processors, providing Intel vRAN Boost technology and up to 64 cores of computing prowess. It features 1x OCP 3.0 slot and 4x PCIe expansion slots capable of accommodating multiple NVIDIA GPU and Smart NIC cards, including NVIDIA L40S, NVIDIA L4 GPU cards, and Bluefield 3/ConnectX-7 NIC Smart NIC Cards. It is ideal for applications in smart cities, intelligent transportation, smart healthcare, retail intelligence, video transcoding, and industrial automation.

Featuring front-access I/O ports, a paint-free design, TPM 2.0 hardware security, and IPMI remote management, the ECA-6040 represents a high-performance platform for building and deploying open, efficient, and secure edge AI services over 5G private networks. Other outstanding features include 4x 2.5” HDD/SSD trays, dual M.2 onboard slots, 1,600W AC redundant PSUs, and 6x swappable smart fans for optimal thermal control.

Results

With its support for multiple GPU and Smart NIC, the ECA-6040 is capable of handling the most demanding AI workloads simultaneously so that all AI applications perform their data crunching locally, resulting in quicker response times and reduced reliance on cloud services.

By enabling a single hardware platform, such as Lanner’s ECA-6040, to incorporate AI alongside other functionalities, multifunction edge devices can run multiple concurrent AI workloads with limited resources through resource partitioning, isolation and remote management.

Equipped with Intel vRAN Boost technology built into the latest 4th Gen Intel Xeon Scalable Processor, the ECA-6040 delivers unparalleled performance and power efficiency for RAN virtualization, ensuring optimal resource utilization in 5G O-RAN deployments.

Conclusion

Through the deployment of an ideal workload consolidation platform such as the ECA-6040, businesses significantly reduce both development time and total costs of operations. The ECA-6040’s specifications and configurations ensure hiccup-free software integration and scalable hardware expansion, ensuring compatibility with existing IT infrastructure. Furthermore, by consolidating AI with 5G using Lanner’s ECA-6040, hardware footprint and associated costs are also significantly reduced.

Featured Product