Workstation 博客
Networking
标记为 Networking 的文章。
NVIDIA Minerva for AI Factories: Turning Alert Storms into Recovered GPU Hours
One degrading optic can raise thousands of alerts and stall a thousand GPUs. What NVIDIA Minerva does, what it is worth in GPU hours, goodput, MTBI, MTTR and tokens, which of its three consumption models fits, worked examples for a Vera Rubin NVL72 neocloud, a university Blackwell cluster and an enterprise with Grafana and ServiceNow, and how it relates to UFM.
Balinder Walia阅读 
NVIDIA UFM for AI Clusters: Why InfiniBand Fabric Visibility Decides What Your GPUs Are Worth
One flapping InfiniBand link can stall a 256-GPU job. What NVIDIA Unified Fabric Manager does, when UFM Telemetry, Enterprise or Cyber-AI fits, worked examples for a 4-node DGX B300 cluster, a 32-node SuperPOD-style cluster and a university Slurm service, and what fabric visibility is worth in GPU hours.
Balinder Walia阅读 
Gateway API Replaces Ingress: A Step-by-Step Migration Playbook
Ingress NGINX is retired and Ingress is frozen. This playbook maps Ingress fields and annotations to Gateway and HTTPRoute, flags the ingress-nginx behaviours that break, and walks through conversion, side-by-side testing, weighted cutover and rollback.
Decoupling Network Policy in Kubernetes: One Owner per Rule
Network rules scattered across annotations, app config and stray NetworkPolicies slow every change. Here is how to split them into layers with one owner each, using ClusterNetworkPolicy, NetworkPolicy and Gateway API, plus a step-by-step migration playbook.
Gateway API Shifts Cluster Networking: An Ownership Model for the Edge
Ingress is frozen and Ingress NGINX is retired. Gateway API replaces one shared object with resources owned by platform and application teams, joined by explicit handshakes. Here is what that means for platform design, RBAC, topology and CRD ownership.