Advanced Computing in the Age of AI | Monday, June 24, 2024

NVIDIA and Cisco Weave Fabric for Generative AI 

June 6, 2024 -- Building and deploying AI applications at scale requires a new class of computing infrastructure — one that can handle the massive amounts of data, compute power and networking bandwidth needed by generative AI models. To better ensure these models perform optimally and efficiently, NVIDIA is teaming with Cisco to enable enterprise generative AI infrastructure.

Cisco’s new Nexus HyperFabric AI cluster solution, developed in collaboration with NVIDIA, provides a path for enterprises to operationalize generative AI. Cisco HyperFabric is an enterprise-ready, end-to-end infrastructure solution to scale generative AI workloads. It combines NVIDIA accelerated computing and AI software with Cisco AI-native networking and a robust VAST Data Platform.

“Enterprise applications are transforming into generative AI applications, significantly increasing data processing requirements and overall infrastructure complexity,” said Kevin Wollenweber, senior vice president and general manager of data center and provider connectivity at Cisco. “Together, Cisco and NVIDIA are advancing HyperFabric to advance generative AI for the world’s enterprises so they can use their data and domain expertise to transform productivity and insight.”

Powering an Enterprise-Ready AI Cluster Solution

Foundational to the solution are NVIDIA Tensor Core GPUs, which provide the accelerated computing needed to process massive datasets. The solution utilizes NVIDIA AI Enterprise, a cloud-native software platform that acts as the operating system for enterprise AI. NVIDIA AI Enterprise streamlines the development and deployment of production-grade AI copilots and other generative AI applications, ensuring optimized performance, security and application programming interface stability.

Included with NVIDIA AI Enterprise, NVIDIA NIM inference microservices accelerate the deployment of foundation models while ensuring data security. NIM microservices are designed to bridge the gap between complex AI development and enterprise operational needs. As organizations across various industries embark on their AI journeys, the combination of NVIDIA NIM and the Cisco Nexus HyperFabric AI cluster supports the entire process, from ideation to development and deployment of production-scale AI applications.

The Cisco Nexus HyperFabric AI cluster solution integrates NVIDIA Tensor Core GPUs and NVIDIA BlueField-3 SuperNICs and DPUs to enhance system performance and security. The SuperNICs offer advanced network capabilities, ensuring seamless, high-speed connectivity across the infrastructure. BlueField-3 DPUs offload, accelerate and isolate the infrastructure services, creating a more efficient AI solution.

BlueField-3 DPUs can also run security services like the Cisco Hypershield solution. It enables an AI-native, hyperdistributed security architecture, where security shifts closer to the workloads needing protection. Cisco Hypershield is another notable area of collaboration between the companies, focusing on creating AI-powered security solutions.

Source: Kevin Deierling, NVIDIA