In Part 2 of our blog series, we close the critical lateral security gap that exists inside the k8s cluster, which is not solved by standard perimeter security. We detail how to use VMware vDefend Distributed Firewall, enforced through Antrea CNI, to implement Zero Trust microsegmentation for GPU-accelerated AI workloads. Specifically, we walk through defining identity-based firewall rules as Terraform code for the NVIDIA NIM RAG Blueprint and demonstrate how Antrea Egress preserves pod identity to block unauthorized access to sensitive external assets, like the Elasticsearch vector database, using packet-by-packet Antrea Traceflow verification
The post From Infrastructure to Agents: A Hands-On Guide to Secure Private AI with Broadcom – Part 2 appeared first on VMware Cloud Foundation (VCF) Blog.