How is NVIDIA EGX shaping the "central nervous system" of industries?

Major change across all industries is being driven by the adoption of blooming IoT sensors across the globe, including cameras for observation, microphones for listening and a host of other smart devices that help businesses. aware and understand what is happening in the real world.

The amount of data generated at the edge is growing exponentially. The only way to handle this massive data in real time is by placing servers near the action point and by harnessing the immense computing power of the GPU.

In the future, a business data center will not have 10,000 servers in one location, but one or more servers in 10,000 different locations. They will be located in office buildings, factories, warehouses, mobile towers, schools, shops and banks. They will detect traffic jams and forest fires, route safe traffic and prevent crime.

By placing a network of distributed servers, where data is transmitted from hundreds of sensors, businesses can use the network settings of data centers at the edge to trigger immediate responses to AI support. In addition, by processing data at the edge, privacy concerns are mitigated and data ownership issues are in place.

Edge servers lack the physical security infrastructure provided by enterprise IT. And companies lack the budget to invest in IT staff specializing in field trips to manage these remote systems. Therefore, edge servers need to be designed to be secure and easy to update, manage and deploy remotely.

In addition, AI systems need to be operational at all times, with zero downtime.

NVIDIA has built the NVIDIA EGX Edge AI platform to ensure global security and resilience. By simplifying deployment and management, NVIDIA EGX enables always-on AI applications to automate important infrastructure in the future. The platform is a Kubernetes-based software platform and GPU-accelerated AI-powered container to everything from dual-socket x86 servers to ARM-based NVIDIA Jetson SoCs.

To date, more than 20 server vendors build edge and micro-edge servers supported by EGX, including ADLINK, Eclech, Atos, AVerMedia, Cisco, Connect Tech, Dell Technologies, Diamond Systems, Fujitsu. , Gigabyte, Hewlett Packard Enterprise, Inspur, Lenovo, Quanta Technologies and Supermicro. As well as dozens of network security and hybrid cloud partners in the NVIDIA edge ecosystem, such as Canonical, Check Point, Excelero, Guardicore, IBM, Nuage Networks, Nutanix, Palo Alto Networks, Rancher, Red Hat, VMware, Weka and Wind River.

There are also hundreds of AI applications and integrated solution providers built on NVIDIA EGX to provide industry-specific services to businesses around the globe.

Businesses that operate the AI ​​platform need to protect not only customer data, but also AI models that help convert data into action. By combining NVIDIA Mellanox SmartNIC, the industry standard for connecting high-performance, secure networks, with AI processors into NVIDIA EGX A100, a combined accelerated convergence processor, NVIDIA says it will introduce Introducing basic new enhancements to AI at the edge.

Enhanced security and performance
Secure, validated boot for GPU and SmartNIC from “root-of-trust” at hardware level ensures the device's firmware and life cycle are safely managed. Third-generation Tensor Core technology in NVIDIA Ampere architecture delivers industry-leading AI performance. Specifically with the EGX A100, the new GPU security tool helps load encrypted AI models and encrypts all output from AI, preventing threats of data theft.

As edge computing gradually encodes high-resolution sensors, SmartNIC supports inline cryptographic acceleration at line speeds. This allows the encrypted feed to be decoded and sent directly to the GPU memory, bypassing the CPU and system memory.

Edges also require a higher level of security to guard against threats from other devices on the network. With the ability to reduce the load on the reconfigurable firewall in hardware, SmartNIC effectively provides the first line of defense for the hybrid cloud, ensuring a secure service network connection.

NVIDIA Mellanox's time-triggered transport technology for telco (5T for 5G) ensures commercial solutions that can meet the most time-sensitive use cases for 5G vRAN with the NVIDIA Aerial SDK . This will lead to a new wave of CloudRAN in the telecommunications industry.

With NVIDIA Ampere GPU and Mellanox ConnectX-6 D on a converged product, EGX A100 provides high throughput, low latency packet handling for virtual network security and functionality.

Simple to deploy, manage and secure at scale
Through NGC, NVIDIA's GPU-optimized container portfolio, we provide industry-standard application frameworks and specialized AI toolset to simplify the start and adjustment of AI applications to the environment. new border. They can be used together or individually and open up new possibilities for many frontier use cases.

And with NGC Private Registry, applications can be signed before publishing to ensure they are not tampered with during transit, then verified before running at the edge. NGC Private Registry also supports model version and encryption, so light model updates can be sent quickly and safely.

The future of edge computing requires secure, scalable, flexible, and manageable systems of AI-enabled systems operating on the edge of the network. By combining the combined acceleration of the NVIDIA GPU and NVIDIA Mellanox SmartNIC together with NVIDIA EGX, NVIDIA is building both platforms and ecosystems to form the AI ​​"nervous system" of all industries around the world. bridge.