Cloudian Next-Gen AI Storage with NVIDIA and Lenovo Reference Architecture

Cloudian, Inc. announced an enterprise storage solution for AI that combines HyperStore object storage with Lenovo Group Ltd.’s latest AI-optimized servers in a NVIDIA-validated architecture.

The platform delivers throughput of 20GB/s per node with linear scaling capabilities, meeting the intense I/O demands of modern AI workloads including large language model training, inference operations, and checkpointing functions.

As organizations increasingly rely on machine learning and deep learning models, object storage has emerged as a foundational technology for AI infrastructure. The company leads this evolution with full S3 API compatibility, enterprise-grade security features, and enhanced performance through NVIDIA GPUDirect Storage (GDS) integration, enabling organizations to maximize their AI investments.

Validated architecture demonstrates key capabilities:

• Flash-optimized performance utilizing NVIDIA GPUDirect Storage and RDMA networking with linear throughput scaling
• Enterprise-class features including observability, self-healing capabilities, multi-tenant support, QoS controls, dynamic storage provisioning, and data lifecycle management
• Comprehensive security framework with enterprise-grade certifications, built-in ransomware protection, and standard OS kernel implementation

“Today’s AI storage requirements demand both extreme performance and enterprise-grade data management,” said Amit Rawlani, Sr. director, alliances, Cloudian. “Our solution transforms AI infrastructure by combining HPC-level performance with the scalability and efficiency of S3-compatible object storage, while delivering the enterprise features organizations require.”

“This validated architecture unites Lenovo’s GPU-optimized systems and flash-optimized storage infrastructure with Cloudian’s enterprise object storage to create a comprehensive AI infrastructure solution,” said Stuart McRae, GM, data storage, Lenovo. “This solution is well suited for the dynamic workload requirements of modern enterprises.”

The reference architecture enables organizations to move quickly from proof-of-concept to enterprise-wide AI deployments. By delivering native S3 compatibility, enterprise-grade security, and a scalable architecture, the platform adapts to growing AI requirements while maintaining consistent performance and centralized management. Organizations can start with a targeted deployment for specific workloads, then expand capacity and performance as business needs evolve.

Resource:
Access reference architecture document (registration required)