Unleash the Power of AI for the Data Awakening
AI-Ready Data Infrastructure Reference Architecture White Paper
This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>
-
Enterprise
Enterprise products, solutions & services
-
Huawei Cloud
-
Carrier
-
Consumer
-
Corporate
Huawei Global - English
-
The era of AI is here. Large AI models are already injecting new vitality into data, further unlocking its inherent value. Indeed, powered by AI, data is becoming something akin to a new oil: the key element for driving new productivity. Data infrastructure, of course, carries data, so the strength of a country's data infrastructure will ultimately determine its ability to move forward and stay ahead.
Chen GuoliangAcademician of the Chinese Academy of Sciences -
Data infrastructure needs to offer high-performance, strong consistency, resilience, and reliable data access services to ensure that high-quality data can be effectively used in AI computing. So, data infrastructure featuring future-ready storage power that’s a perfect fit for AI computing has become the key to bolstering the era of large AI models.
Dr. Peter ZhouPresident of the IT Product Line, Huawei
Challenges for Data Infrastructure in the AI Era
Data Asset Management
Enterprises need diverse capabilities in order to deal with data of varying quality, drawn from numerous sources, support full data mobility between different departments.
Cluster
Utilization
Utilization
The computing power utilization of a large-scale training cluster is less than 50%. This results in high computing construction costs and high power consumption.
Data
Consistency
Consistency
Any user must be able to access the latest updated data copy at any time, on any node.
Data
Resilience
Resilience
AI-based applications lead to new vulnerabilities, endangering resilience. There is potential here for a loss of tens of millions of dollars after a large AI model is attacked.
Definition and Characteristics of AI-Ready Data Infrastructure
AI-ready data infrastructure refers to data storage software and hardware systems designed for AI applications and services. Building AI-ready data infrastructure therefore requires comprehensive preparation, across the board.
Reference Architecture
Central AI clusters can now have tens of thousands — if not hundreds of thousands — of cards. With edge AI models beginning to permeate many industries at a faster pace, Huawei’s AI-ready data infrastructure reference architecture and solution help enterprises across different industries build resilient, reliable, and open AI data infrastructure, empowering them to elevate their intelligence and innovation to the next level.
Advice for CIOs
-
Build Unified Data Lakes
AI is only as good as the data that fuels it. Enterprises must build unified data lakes to enable data asset visibility, manageability, and availability, turn data into a key element of production, and implement large AI model services faster.
-
Employ AI-Ready Data Infrastructure
Ever-evolving large AI models have increasingly higher computing power needs. Enterprises should refer to industry best practices to shift from simply stacking computing power to fully unlocking its potential. This requires enterprises to build a future-proof intelligent computing foundation powered by an AI-ready data infrastructure solution.
-
Embed All-Round Data Protection Capabilities
Data assets have become increasingly critical to large AI models. Tampering, theft, and ransomware attacks to core data such as model files and training data can cause massive resource waste and economic losses. Enterprises should immediately build all-round data protection capabilities.
-
Accelerate AI Adoption via HCI Appliances
Large AI model training for data center applications requires huge investments. Edge applications are a key area where generative AI can create profits. Enterprises should use one-stop training/inference HCI appliances to quickly develop and launch their AI products to monetize their large AI models.
-
Build Specialized Technical Teams
Enterprises should proactively evaluate their AI readiness, build technical teams specialized in large AI models (especially storage for large AI models), and improve their professional capabilities.