AI in Aruba
- AI in Aruba: Server Configuration & Deployment
This article details the server configuration used to support Artificial Intelligence (AI) workloads within the Aruba Networks environment. This is intended as a guide for new systems administrators and engineers deploying or maintaining AI-related infrastructure.
Overview
The “AI in Aruba” initiative leverages a hybrid server architecture, combining on-premise hardware for low-latency processing with cloud-based resources for model training and large dataset storage. This allows for real-time insights from network data while maintaining scalability and cost-effectiveness. This document focuses on the on-premise server configuration, which forms the core of the real-time analytics pipeline. We utilize a combination of high-performance compute servers and dedicated storage arrays. Understanding the interplay between Network Infrastructure and server performance is crucial.
Hardware Specifications
The core processing is handled by dedicated GPU servers. The following table outlines the key specifications for these servers:
| Component | Specification | Quantity per Server |
|---|---|---|
| CPU | Dual Intel Xeon Gold 6338 (32 cores/64 threads per CPU) | 2 |
| RAM | 512 GB DDR4 ECC Registered 3200MHz | 1 |
| GPU | NVIDIA A100 80GB PCIe 4.0 | 4 |
| Storage (OS) | 500GB NVMe PCIe Gen4 SSD | 1 |
| Storage (Data) | 4TB NVMe PCIe Gen4 SSD (RAID 0) | 1 |
| Network Interface | Dual 100GbE QSFP28 | 1 |
| Power Supply | 2000W Redundant Platinum | 2 |
These servers are housed in a dedicated, climate-controlled server room with redundant power and cooling. Proper Server Room Management is essential for maintaining stability. See also Power Distribution Units for details on power infrastructure.
Storage Configuration
Data storage is a critical component, requiring both high capacity and low latency. We employ a tiered storage approach. Raw data is stored in a cloud-based object storage solution (AWS S3, Google Cloud Storage), while frequently accessed data is cached on-premise.
The on-premise caching layer utilizes a dedicated NVMe storage array. Details are below:
| Parameter | Value |
|---|---|
| Vendor | Pure Storage |
| Model | FlashArray//X70 |
| Capacity | 100TB Raw |
| Usable Capacity | 50TB (after RAID and Deduplication) |
| RAID Level | RAID-TP (Triple Parity) |
| Connectivity | 100GbE iSCSI |
| Data Reduction | Inline Deduplication & Compression |
This array provides the necessary I/O performance for real-time data analysis. Understanding Storage Area Networks is crucial for managing this infrastructure. We also employ Data Backup Strategies to protect against data loss.
Software Stack
The software stack is built on a Linux foundation, providing flexibility and control.
| Component | Version | |
|---|---|---|
| Operating System | Ubuntu Server 22.04 LTS | |
| Containerization | Docker 24.0.5 | |
| Orchestration | Kubernetes 1.27 | |
| AI Framework | TensorFlow 2.13.0 | |
| Programming Language | Python 3.10 | |
| Data Processing | Apache Spark 3.4.1 | |
| Monitoring | Prometheus & Grafana |
All AI models are containerized using Docker and orchestrated with Kubernetes for scalability and resilience. Kubernetes Deployment procedures should be followed carefully. We utilize Continuous Integration/Continuous Deployment (CI/CD) pipelines for automated model updates. Refer to Security Best Practices for securing the software stack. The Networking Configuration must allow for communication between components.
Network Considerations
High bandwidth and low latency are essential for AI workloads. The servers are connected to the Aruba Networks core network via dual 100GbE uplinks. VLAN Segmentation is used to isolate AI traffic from other network traffic. Quality of Service (QoS) policies are implemented to prioritize AI-related data flows. The Firewall Configuration is configured to allow only necessary traffic to and from the AI servers. Load Balancing is employed to distribute traffic across multiple servers.
Future Expansion
Planned future expansion includes:
- Adding more GPU servers to increase processing capacity.
- Implementing a dedicated NVMe-oF (NVMe over Fabrics) network for even lower latency storage access.
- Integrating with additional data sources.
- Exploring the use of more advanced AI frameworks like PyTorch.
Refer to Scalability Planning for long-term infrastructure growth.
Server Maintenance is critical to long-term stability.
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.* ⚠️