A two-node multi-user AI workgroup for clinical research, production intelligence, training and resilient service placement.
This brief explains where the baseline fits, the software and operating model it supports, and the evidence Axiotech should use to validate a final configuration. Indicative specifications remain subject to component availability, export compliance and formal quotation.
Baseline architecture
Two 4U nodes; four NVIDIA RTX PRO 6000 96GB GPUs; 100GbE RDMA; shared protected storage; KVM and Kubernetes-ready infrastructure.
The current compute node is based on the Supermicro AS-4125GS-TNRT / CSE-418G2TS 4U rack platform, integrated with matched rails, power, management, networking and rack infrastructure as required.
Best-fit workloads
Two-node medical-research AI, production intelligence, multi-user training, resilient inference placement and a small private AI platform with four 96GB GPUs.
It separates service and training workloads across physical nodes while still supporting distributed GPU jobs over RDMA.
Recommended software stack
Kubernetes with NVIDIA GPU Operator/device plugin or Slurm, CUDA/NCCL, PyTorch, Triton/vLLM, MLflow, Prometheus/Grafana, registry and secrets management.
Pin host drivers and infrastructure separately from versioned application containers. Record source, image, dataset/model and hardware allocation with each benchmark or production release.
Deployment pattern
Use Kubernetes or Slurm with explicit GPU allocation, central identity, a protected registry and shared storage. Separate validation/training and operational services by node or scheduled policy.
Define monitoring, identity, backup, change control and workload ownership at the same time as compute. Multi-user platforms require resource allocation and quotas; production services require health, overload and rollback behaviour.
Sizing boundary
Cross-node training depends on network, NCCL and topology configuration. Regulated workloads also require an application-level validation and governance plan.
Final sizing should use representative code, data, concurrency and service objectives. Aggregate core, RAM or VRAM figures do not by themselves predict application performance.
Commissioning and acceptance
Test single- and cross-node training, node loss and service rescheduling, image/model rollback, shared-storage recovery and the data-governance controls required by the intended workload.
Axiotech should retain the resulting configuration, firmware/driver baseline, environment manifest, benchmark data and recovery procedure as the system acceptance pack.
Primary technical references
- NVIDIA Container Toolkit overview
- Prometheus overview
- Slurm Quick Start Administrator Guide
- NVIDIA CUDA C++ Best Practices Guide
- NVIDIA NCCL documentation
References are provided for software architecture and implementation planning. Validate the versions, licences, support matrix and regulated-use requirements applicable to the final deployment.