OceanStor AI Storage
Build a solid data foundation for AI, placing intelligence at your fingertips.
This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>
This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>
Enterprise products, solutions & services
Build a solid data foundation for AI, placing intelligence at your fingertips.
Designed for AI
As large AI models move into the multimodal era, enterprises face challenges in model training and inference, from low training computing power utilization, slow inference response, and hallucinations, to an inability to handle long sequences and high inference costs.
And the answer? Huawei OceanStor Next-Gen High-Performance Distributed File Storage for AI provides a unified storage solution for the end-to-end (E2E) AI training and inference data process. It helps enterprises overcome data silos, aggregate diverse corpus data, improve AI cluster computing power utilization, and enhance the inference experience.
In the globally recognized MLPerf benchmark test, OceanStor A series storage ranked first in performance for consecutive years, with training set loading 8x faster and training resumption from checkpoints 4x faster than the leading alternative.
Six Innovations
In response to the demands of large AI model training and inference, Huawei redefines distributed file storage, focusing on six key innovations to create storage systems purpose-built for AI.
Ultra Performance
Storage performance is improved 10-fold compared to conventional storage, with bandwidth in hundreds of TB/s and 100 million-level input/output operations per second (IOPS), speeding up the entire Generative AI (GenAI) process.New Data Paradigm
Cutting-edge AI data paradigms are supported, including tensors, vectors, and Key-Value (KV) cache. Retrieval-Augmented Generation (RAG) technology reduces AI hallucinations. Multi-level KV cache reduces the Time-To-First-Token (TTFT) while improving inference efficiency.Data Fabric
Storage metadata management and retrieval help achieve global data visibility and manageability, as well as data mobility that is 10 times more efficient.Scalability
Exabyte-level scale-out of a single storage cluster is supported, with each controller enclosure scaling up with Graphics Processing Units (GPUs), Data Processing Units (DPUs), or Neural Processing Units (NPUs) for near-storage computing.Data Resilience
Innovative architecture and technology achieve 99.9999% reliability. A built-in ransomware detection engine delivers 99.99% detection accuracy.Sustainability
Storage media application and system hardware innovation significantly improve storage energy efficiency and capacity density.A Product That's Right for You
Using data-control plane separation architecture and long-term memory storage, this storage system fulfills the E2E data processing needs for AI training and inference in various industry sectors, such as financial credit, investment research, healthcare, and drug development.
OceanStor A600 provides extreme performance density, accelerates inference, and fulfills the E2E data processing needs for AI training and inference. As such, it can be widely used in industry-specific scenarios such as financial investment and research, legal documents, medical record review, and drug R&D.
Solution
Huawei AI Data Lake Solution delivers a high-quality AI corpus and speeds up model training and inference, empowering enterprises to embrace AI.
For Performance in MLPerf AI Benchmarks for Consecutive Years
Resources
Load More
Load More
Load More
You Might Be Interested
What is Storage for AI?
Storage for AI is the data storage infrastructure purpose-built for AI and Machine Learning (ML) workloads. It needs to meet the demands of high performance and scalability in hybrid workload environments, effectively manage the massive amounts of data required throughout the AI process, and ensure rapid data reads, writes, and processing. Storage systems designed for AI help training clusters improve the efficiency of data ingestion and preprocessing, increasing the utilization of computing clusters. They also enhance the cost-effectiveness and accuracy of inference applications.
Copyright © 2025 Huawei Technologies Co., Ltd. All rights reserved.