Huawei Atlas 900 PoD - User Manual

To Next Page

1 2 3

Atlas 900 PoD Quick Start Guide

(Model 9000, AC)

Document Issue: 04 (2023-03-30) BOM: 31500ESN (2/2)

Huawei Technologies Co., Ltd.

No part of this document may be reproduced or transmitted in any form or by any means without prior written

consent of Huawei Technologies Co., Ltd.

and other Huawei trademarks are trademarks of Huawei Technologies Co., Ltd.

All other trademarks and trade names mentioned in this document are the property of their respective holders.

The purchased products, services and features are stipulated by the contract made between Huawei and the

customer. All or part of the products, services and features described in this document may not be within the

purchase scope or the usage scope. Unless otherwise specified in the contract, all statements, information, and

recommendations in this document are provided "AS IS" without warranties, guarantees or representations of any

kind, either express or implied.

The information in this document is subject to change without notice. Every effort has been made in the

preparation of this document to ensure accuracy of the contents, but all statements, information, and

recommendations in this document do not constitute a warranty of any kind, e

press or implied.

Symbol

Description

Indicates a hazard with a high level of risk which, if not avoided, will

result in death or serious injury.

Indicates a hazard with a medium level of risk which, if not avoided,

could result in death or serious injury.

Indicates a hazard with a low level of risk which, if not avoided, could

result in minor or moderate injury.

Indicates a potentially hazardous situation which, if not avoided,

could result in equipment damage, data loss, performance

deterioration, or unanticipated results.

NOTICE is used to address practices not related to personal injury.

Supplements the important information in the main text.

NOTE is used to address information not related to personal injury,

equipment damage, and environment deterioration.

1 Specifications

The Atlas 900 PoD (model 9000) is a basic AI cluster unit launched by Huawei. It features high

density, high performance, energy efficiency, reliability, flexible scalability, cabinet-based

delivery, simplified O&M, and low TCO.

Environmental specifications:

Item

Specification

Temperature

⚫ Operating range: 5°C to 40°C (41°F to 104°F) (Operating

altitude ≤ 1800 m, ASHRAE Classes A2, A3, and A4 compliant)

⚫ Storage range (within three months): –30°C to +60°C (–22°F to

+140°F)

⚫ Storage range (within six months): –15°C to +45°C (5°F to

113°F)

⚫ Maximum change rate: 20°C/h (36°F/h), 5°C (9°F) per 15

minutes

Relative humidity

(RH, non-condensing)

⚫ Operating range: 8% to 90%

⚫ Storage humidity (within 72 hours): 8% to 95%

⚫ Storage humidity (within six months): 20% to 75%

Altitude

Operating altitude: ≤ 3,050 m (10,006.56 ft)

ASHRAE 2015 compliant:

⚫ When the server configuration complies with ASHRAE Classes

A1 and A2 and the altitude is above 900 m (2952.76 ft), the

highest operating temperature decreases by 1°C (1.8°F) for

every increase of 300 m (984.24 ft).

⚫ When the configuration complies with ASHRAE Class A3

standards and the altitude is above 900 m (2952.76 ft.), the

highest operating temperature decreases by 1°C (1.8°F) for

every increase of 175 m (574.14 ft.).

⚫ When the configuration complies with ASHRAE Class A4

standards and the altitude is above 900 m (2952.76 ft.), the

highest operating temperature decreases by 1°C (1.8°F) for

every increase of 125 m (410.1 ft.).

Load-bearing

requirements for the

equipment room

Load-bearing capacity > (Total weight of cabinets + Total weight

of maintenance personnel)/Floor area occupied by cabinets

Where:

⚫ It is assumed that each cabinet is maintained by one person in

the front and one in the rear of the cabinet and that the

weight of each person is 100 kg (220.46 lb.).

⚫ Floor area occupied by cabinets = Cabinet width x (Cabinet

depth + Front aisle width/2 + Rear aisle width/2)

A schematic diagram of the external structure:

The water inlet is blue and the water outlet is red.

2 Tools

⚫

Paper knife: Used to cut the sealing tape of the packing case.

⚫

Phillips screwdriver: Used to tighten small screws.

⚫

Electrostatic discharge (ESD) gloves: Used to prevent ESD damage when you handle a server

component or precision instrument.

⚫

Adjustable wrench: Used to adjust the leveling feet of the cabinet.

3 Installation

⚫ Exercise caution when using tools.

⚫ If the installation position of the device is higher than the shoulders of the installation

personnel, use a vehicle such as a lift to facilitate installation. Prevent the equipment from

falling down and causing personal injury or damage to the equipment. When a ladder is

used, ensure that another person holds the ladder steady to prevent accidents.

⚫

Wear ESD gloves and remove conductive objects, such as jewelry and watches.

⚫

At least four persons are required to move the device. When moving a device, keep your

back straight and move smoothly to avoid any injury.

⚫

Do not use mounting ears as the holding points when moving the device. Otherwise, the

device may be damaged or slipped.

1. Unpack the cabinet.

Liquid-cooled cabinet

Air/liquid heat exchanger

Power module

Management plane switch (example)

Compute node

Parameter plane switch (example)

Service plane switch (example)

Need help?

Do you have a question about the Huawei Atlas 900 PoD and is the answer not in the manual?

The Atlas 900 PoD (model 9000) is a basic AI cluster unit developed by Huawei, designed for high-density, high-performance computing environments. This system emphasizes energy efficiency, reliability, and flexible scalability, making it suitable for various AI workloads. It is delivered as a cabinet-based solution, which simplifies operations and maintenance (O&M) and contributes to a lower total cost of ownership (TCO).

Function Description:

The Atlas 900 PoD serves as a foundational AI cluster unit, integrating multiple components within a single cabinet to provide a complete computing solution. It is designed to handle demanding AI tasks by offering high-density compute nodes, efficient cooling, and robust power delivery. The system's architecture supports the deployment of various AI applications, leveraging its high performance to accelerate training and inference processes. The cabinet-based delivery simplifies deployment, as core components are pre-integrated and configured.

Important Technical Specifications:

Environmental Specifications:

Temperature:
- Operating Range: 5°C to 40°C (41°F to 104°F) when the operating altitude is ≤ 1800 m, compliant with ASHRAE Classes A2, A3, and A4.
- Storage Range (within three months): -30°C to +60°C (-22°F to +140°F).
- Storage Range (within six months): -15°C to +45°C (5°F to 113°F).
- Maximum Change Rate: 20°C/h (36°F/h), or 5°C (9°F) per 15 minutes.
Relative Humidity (RH, non-condensing):
- Operating Range: 8% to 90%.
- Storage Humidity (within 72 hours): 8% to 95%.
- Storage Humidity (within six months): 20% to 75%.
Operating Altitude: ≤ 3,050 m (10,006.56 ft).
ASHRAE 2015 Compliance Notes for Altitude:
- When the server configuration complies with ASHRAE Classes A1 and A2, and the altitude is above 900 m (2952.76 ft), the highest operating temperature decreases by 1°C (1.8°F) for every increase of 300 m (984.24 ft).
- When the configuration complies with ASHRAE Class A3 standards, and the altitude is above 900 m (2952.76 ft), the highest operating temperature decreases by 1°C (1.8°F) for every increase of 175 m (574.14 ft).
- When the configuration complies with ASHRAE Class A4 standards, and the altitude is above 900 m (2952.76 ft), the highest operating temperature decreases by 1°C (1.8°F) for every increase of 125 m (410.1 ft).
Load-bearing Requirements for the Equipment Room:
- Load-bearing capacity must be greater than (Total weight of cabinets + Total weight of maintenance personnel) / Floor area occupied by cabinets.
- It is assumed that each cabinet is maintained by one person at the front and one at the rear, with each person weighing 100 kg (220.46 lb).
- Floor area occupied by cabinets is calculated as: Cabinet width x (Cabinet depth + Front aisle width/2 + Rear aisle width/2).

Usage Features:

Cabinet-based Delivery: The system is delivered as a pre-integrated cabinet, simplifying the installation process and reducing deployment time. This approach ensures that all necessary components are properly installed and configured before reaching the customer site.
High Density and Performance: Designed for demanding AI workloads, the Atlas 900 PoD offers high computing density, allowing for efficient utilization of space and resources. Its high performance capabilities are crucial for accelerating AI model training and inference.
Energy Efficiency: The system incorporates features for energy efficiency, including liquid cooling, to minimize power consumption and operational costs. The liquid-cooled cabinet (1) and air/liquid heat exchanger (2) are key components in achieving this.
Flexible Scalability: The modular design allows for flexible scalability, enabling users to expand their AI infrastructure as their needs grow. This ensures that the system can adapt to evolving requirements without significant overhauls.
Simplified O&M: The cabinet-based design and integrated management features contribute to simplified operations and maintenance. Components like the management plane switch (4) and parameter plane switch (6) facilitate easy monitoring and control.
Liquid Cooling System: The device utilizes a liquid cooling system, indicated by the liquid-cooled cabinet (1) and air/liquid heat exchanger (2). The water inlet is blue, and the water outlet is red, providing a clear visual distinction for connection.
Power-On Procedure: A structured power-on sequence ensures safe and correct system startup. This includes verifying PSU, management module, compute node, and switch installation, connecting power cables, and ensuring proper grounding. The system requires the CDU (Cooling Distribution Unit) to be powered on and running properly before the liquid-cooled cabinet is powered on.
Hazardous Substance Compliance: The product complies with SJ/T 11364 standards for hazardous substances in electronic and electrical products, indicating a commitment to environmental responsibility.

Maintenance Features:

Component Accessibility: The design allows for access to internal components for maintenance. The schematic diagram shows various modules such as compute nodes (5), power modules (3), and service plane switches (7), which are likely designed for easy replacement or servicing.
Tool Requirements: Standard tools are required for installation and maintenance, including a paper knife for unpacking, a Phillips screwdriver for small screws, ESD gloves for handling sensitive components, and an adjustable wrench for leveling feet.
Safety Guidelines: Clear safety warnings and cautions are provided for installation and maintenance, emphasizing the importance of wearing ESD gloves, handling heavy devices with sufficient personnel, and exercising caution when using tools. This ensures the safety of personnel and prevents damage to the equipment.
Grounding: Proper grounding of all modules is critical for safety, as the equipment is powered by high-voltage sources. This is a key maintenance consideration to prevent serious injury or death from electrical shock.
Visual Indicators: Power indicators on BBUs, management modules, compute nodes, and switches provide visual feedback on the operational status of the modules, aiding in troubleshooting and maintenance.
Unpacking and Setup: The manual provides detailed steps for unpacking the cabinet, including slitting straps, removing covers and cardboards, and securing the sliding plate to the pallet. This ensures a safe and systematic initial setup.
Leveling Feet Adjustment: Adjustable leveling feet are provided to ensure the cabinet is stable and level, which is crucial for proper operation and longevity of the equipment. This adjustment is performed using an adjustable wrench.