DGX Pod


An Accelerated Analytics/AI cluster for the Enterprise

AI Infrastructure for the ENTERPRISE dataCenter

The NVIDIA DGX POD and SuperPOD platforms are based on a reference architecture that utilizes best practices for integrating compute, storage and networking into an performant, scalable, AI infrastructure for deployment to the modern enterprise datacenter.

DeepINSIGHTS DGX POD

- A choice of DGX-1 or DGX-2H for the compute
- Mellanox switching for Infiniband and Ethernet
- A choice of IBM ESS Scale or DDN A3I for Storage
- DGX Pod Management software stack

DGX POD MANAGEMENT

- DGX OS Software based on Ubuntu or RedHat
- Helmed by Kubernetes for AI workloads
- Prometheus for data collection and analysis
- NGC Cloud for highly tuned containers and Model libraries