AI in San Marino
- AI in San Marino: Server Configuration & Deployment
This article details the server configuration used for deploying Artificial Intelligence (AI) applications within the Republic of San Marino. It is aimed at newcomers to the server infrastructure and provides a detailed overview of the hardware, software, and networking components used. This setup prioritizes scalability, security, and performance for demanding AI workloads.
Overview
The AI infrastructure in San Marino is designed to support a variety of applications, including machine learning model training, inference serving, and data analytics. The core of the system is a distributed cluster of high-performance servers, interconnected via a low-latency network. Data storage is handled by a dedicated storage array, ensuring rapid access to large datasets. We utilize a hybrid cloud approach, leveraging both on-premise hardware and cloud resources for flexibility and cost-effectiveness. Access control is managed through a combination of firewalls, intrusion detection systems, and user authentication protocols. See Security Considerations for more detailed information.
Hardware Configuration
The server cluster comprises several node types, each optimized for specific tasks. The following table details the specifications of the main node types:
| Node Type | CPU | RAM | GPU | Storage | Network Interface |
|---|---|---|---|---|---|
| Compute Node (Training) | 2 x AMD EPYC 7763 (64 cores, 128 threads) | 512GB DDR4 ECC REG | 4 x NVIDIA A100 (80GB) | 4 x 4TB NVMe PCIe Gen4 SSD (RAID 0) | 100Gbps InfiniBand |
| Compute Node (Inference) | 2 x Intel Xeon Gold 6338 (32 cores, 64 threads) | 256GB DDR4 ECC REG | 2 x NVIDIA T4 | 2 x 2TB NVMe PCIe Gen4 SSD (RAID 1) | 25Gbps Ethernet |
| Storage Node | 2 x Intel Xeon Silver 4310 (12 cores, 24 threads) | 128GB DDR4 ECC REG | None | 16 x 16TB SAS HDD (RAID 6) | 40Gbps Ethernet |
All servers are housed in a secure data center in Serravalle, San Marino, with redundant power and cooling systems. Power usage effectiveness (PUE) is monitored continuously and optimized for efficiency. See Data Center Infrastructure for more details.
Software Stack
The software stack is built around a Linux distribution (Ubuntu 22.04 LTS) and incorporates several key open-source technologies. Containerization is heavily utilized via Docker and Kubernetes for application deployment and orchestration. Machine learning frameworks such as TensorFlow, PyTorch, and scikit-learn are pre-installed on the compute nodes.
The following table summarizes the core software components:
| Component | Version | Purpose |
|---|---|---|
| Operating System | Ubuntu 22.04 LTS | Base operating system for all servers |
| Containerization | Docker 23.0.6 | Application packaging and isolation |
| Orchestration | Kubernetes 1.27 | Container deployment and management |
| Machine Learning Framework | TensorFlow 2.12 | Deep learning model development and training |
| Machine Learning Framework | PyTorch 2.0.1 | Deep learning model development and training |
| Data Science Library | scikit-learn 1.2.2 | Machine learning algorithms and tools |
Version control is managed using Git, and a CI/CD pipeline is implemented using Jenkins for automated software builds and deployments. Monitoring and logging are handled by Prometheus and Grafana, providing real-time insights into system performance. See Software Deployment Procedures for more information.
Networking Configuration
The server cluster is connected via a high-speed, low-latency network. The network topology is a fat-tree architecture, providing multiple paths between nodes. Inter-node communication is primarily handled via InfiniBand for training nodes and Ethernet for inference and storage nodes.
The following table details the network configuration:
| Network Segment | Technology | Speed | Purpose |
|---|---|---|---|
| Interconnect (Training Nodes) | InfiniBand HDR | 200 Gbps | High-performance communication for distributed training |
| Interconnect (Inference Nodes) | Ethernet 25Gbps | Communication between inference servers | |
| Storage Network | Ethernet 40Gbps | Access to the central storage array | |
| Management Network | Ethernet 1Gbps | Server management and monitoring | |
| External Access | Firewall Protected Ethernet 10Gbps | Secure access to the AI infrastructure from outside the data center. |
Network security is enforced by a combination of firewalls, intrusion detection systems, and access control lists. All network traffic is encrypted using TLS/SSL. See Network Security Protocols for details. We use Virtual Private Clouds for isolation of sensitive data.
Future Enhancements
Planned future enhancements include the integration of specialized AI accelerators, such as Google TPUs, and the expansion of the storage capacity to accommodate growing datasets. We also plan to explore the use of federated learning techniques to enable collaborative model training without sharing sensitive data. Further improvements to the Monitoring and Alerting Systems are also planned.
Server Maintenance Schedule provides information about planned downtime.
Contact Support for assistance.
Intel-Based Server Configurations
| Configuration | Specifications | Benchmark |
|---|---|---|
| Core i7-6700K/7700 Server | 64 GB DDR4, NVMe SSD 2 x 512 GB | CPU Benchmark: 8046 |
| Core i7-8700 Server | 64 GB DDR4, NVMe SSD 2x1 TB | CPU Benchmark: 13124 |
| Core i9-9900K Server | 128 GB DDR4, NVMe SSD 2 x 1 TB | CPU Benchmark: 49969 |
| Core i9-13900 Server (64GB) | 64 GB RAM, 2x2 TB NVMe SSD | |
| Core i9-13900 Server (128GB) | 128 GB RAM, 2x2 TB NVMe SSD | |
| Core i5-13500 Server (64GB) | 64 GB RAM, 2x500 GB NVMe SSD | |
| Core i5-13500 Server (128GB) | 128 GB RAM, 2x500 GB NVMe SSD | |
| Core i5-13500 Workstation | 64 GB DDR5 RAM, 2 NVMe SSD, NVIDIA RTX 4000 |
AMD-Based Server Configurations
| Configuration | Specifications | Benchmark |
|---|---|---|
| Ryzen 5 3600 Server | 64 GB RAM, 2x480 GB NVMe | CPU Benchmark: 17849 |
| Ryzen 7 7700 Server | 64 GB DDR5 RAM, 2x1 TB NVMe | CPU Benchmark: 35224 |
| Ryzen 9 5950X Server | 128 GB RAM, 2x4 TB NVMe | CPU Benchmark: 46045 |
| Ryzen 9 7950X Server | 128 GB DDR5 ECC, 2x2 TB NVMe | CPU Benchmark: 63561 |
| EPYC 7502P Server (128GB/1TB) | 128 GB RAM, 1 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (128GB/2TB) | 128 GB RAM, 2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (128GB/4TB) | 128 GB RAM, 2x2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (256GB/1TB) | 256 GB RAM, 1 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (256GB/4TB) | 256 GB RAM, 2x2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 9454P Server | 256 GB RAM, 2x2 TB NVMe |
Order Your Dedicated Server
Configure and order your ideal server configuration
Need Assistance?
- Telegram: @powervps Servers at a discounted price
⚠️ *Note: All benchmark scores are approximate and may vary based on configuration. Server availability subject to stock.* ⚠️