Building AI/ML Data Center Fabric with VXLAN EVPN

This guide presents a comprehensive design for building data center networks optimized for Artificial Intelligence (AI) and Machine Learning (ML) workloads using Cisco Nexus 9000 Series switches and VXLAN EVPN fabric. The document outlines the requirements for scalable, high-performance networking to support large-scale GPU clusters, detailing the architecture that enables low latency and high bandwidth. It describes the design concept, including rail-optimized mapping and network segmentation for secure multi-tenancy. The guide also includes a customer use case demonstrating the deployment of 88 NVIDIA GPU servers, highlighting key outcomes such as consistent performance, agile operations, and secure environments for research and production. Additionally, it covers the technology overview of VXLAN EVPN and its benefits for efficient traffic management and workload mobility. The document serves as a blueprint for organizations looking to implement robust AI/ML data center networks.