Data Center Infrastructure for AI Systems

Date: October 16-17

Live-Online Training (MST Time) 7:00am - 5:00pm

Course Overview:
This course provides a comprehensive understanding about key elements to consider when designing, managing, and optimizing data center infrastructure specifically to support Artificial Intelligence (AI) workloads. Participants will explore the unique requirements, challenges, and best practices for AI-centric data centers, focusing on power, cooling, sustainability and safety considerations essential for high-performance AI applications.

1.AI Evolution

• AI Evolution Timeline

• Near Future of AI Systems

2. AI, ML, DL, Introduction

• understanding AI Processes

• AI Workloads

  • Training

  • Inference

• HPC Workloads

3. AI Systems Hardware

• AI Servers

  • Type of Processors (CPU, GPU, DPU, TPU, NPU)

    • Features

    • Typical Applications

• AI Storage

• AI Networking

• AI Clustering

4. AI Racks Considerations

• Rack Densities Categories

• Sizing AI Rack Power

  • Design Rack Density

  • Examples of Rack Elevations (42u, 45u, 48u, etc)

  • Weight Considerations

5. Cooling Technologies and Distributions

• Fundamentals

  • Cooling Overview

  • Facility Heat Transfer (FWS/CWS)

  • Environmental Conditions

    • CFM, IAT Temp. RH

    • Dust and Particle Control

  • Cooling: Air-based Vs. Liquid-based

    • Rack Densities and Cooling Technologies

    • Efficiency

    • Reliability

• Air-Based Cooling

  • Airflow Limitations

  • Air Management

    • Containment Strategies

• Liquid Cooling

  • Liquid Cooling Overview

  • Considerations (Water Quality, etc)

  • Rear Door Heat Exchanger

    • Passive Rear Door HX

    • Active Rear Door HX

  • Immersion Cooling (IC)

    • Single-Phase IC

    • Two-Phase IC

  • Direct-to-Chip (D2C)

    • Single-Phase D2C

    • Two-Phase D2C

  • Other Systems

6. Power Distributions

• Data Center Power Distribution Overview

  • DC vs AC Power Distributions

• Other Power / Electrical Considerations

  • Higher Densities and Recovery Time Objectives (RTO)

7. Workshop - Reference Designs

• Case Studies

8. Sustainability and Energy Efficiency

• Global Impact of Data Center Industry

• Controlling Water and Emissions

• Controlling Energy Efficiency

9. Clean On Site Power Generation

• Potential Need for Continuous Power Generation (On Site)

• Clean Energy Sources

  • Hydrogen

  • Biodiesel

  • Nuclear

  • Natural Gas as a “Bridge Fuel”

10. Operation and Management Considerations

• Operations & Management

• Safety Considerations

• Maintainability Considerations

• Installing and Decommissioning ITE

NOTE: Course content is subject to change without notice.

Images Credits:

• Oak Ridge National Laboratory

• 2023 OCP Global Summit / DRJ-En Español 2022 Ibero-America Conference / QNAP Event / SERES, LLC Events

• Rack Densities and Supercomputers & AI Systems - SERES, LLC / 7×24Exchange. Article