GPU-as-a-Service
SNS Network, in strategic partnership with Dell and NVIDIA, is set to launch a fully managed AI cloud infrastructure powered by 64 x NVIDIA H100 GPUs.
This locally hosted GPU-as-a-Service is specifically designed to provide customers with the high-performance computing resources required to support their AI workloads.
Why GPU-as-a-Service (GPUaaS)
Cost Efficiency
Conduct POC on local infrastructure
Scalability and flexibility
Boost ROI by Optimising GPU usage
Enhance performance for AI and Machine Learning (ML)
Sustainability
Why SNS?
Data Sovereignty and Privacy
Scalability and Flexibility
Personalized AI Readiness Assessments
Explore Dell Technologies Extensive Portfolio To Bring AI To Your Data
Infra and Platform Software Stack
| Item | Description | Model | Quantity | Remarks |
|---|---|---|---|---|
| Servers | Compute Nodes | Dell PowerEdge XE9680 | 4 | H100 GPUs |
| Control Plane Nodes | Dell PowerEdge R660 | 3 | No GPUs | |
| Switches | Fabric / Compute Switches | NVIDIA MQM9700 | 2 | InfiniBand switch |
| Storage / Kubernetes Switches | Dell PowerSwitch Z9432F | 2 | Storage and client traffic | |
| Management Switches | Dell PowerSwitch N3248TE-ON | 1 | Management network | |
| Storage | Storage | Dell PowerScale F710 | 3 | Pool usable capacity: 152TB @ 100% utilisation |
| Software | Software | NVIDIA AI Enterprise | 1 | NVIDIA software |
| GPU Orchestration Platform | NVIDIA Platform | 1 | NVIDIA platform |
Subscription Options
We offer flexible subscription options
Monthly
Annually
Reserved instances for 1, 2, or 3 years
How the Technology Works
NVIDIA Multi-Instance GPU (MIG)
Without MIG, different jobs running on the same GPU, such as different AI inference requests, compete for the same resources. A job consuming larger memory bandwidth starves others, resulting in several jobs missing their latency targets. With MIG, jobs run simultaneously on different instances, each with dedicated resources for compute, memory, and memory bandwidth, resulting in predictable performance with QoS and maximum GPU utilisation.
