IBM Storage Scale System 6000 has added certification for the NVIDIA GB200 NVL72 platform, in addition to previously supported NVIDIA HGX™ H100, H200, and B200 systems. As a certified NVIDIA Cloud Partner (NCP), Storage Scale System 6000 has proven it can deliver scalable, low-latency I/O performance to meet the rigorous demands of AI training and inferencing at scale on NVIDIA GPU architectures.
Built for speed and simplicity, Storage Scale System 6000 provides a turnkey platform to power global data infrastructure for enterprise file and object workloads. It merges IBM Storage Scale software with NVMe flash and hybrid storage configurations to support AI pipelines, data analytics, and mission-critical file services.
NVIDIA’s HGX™ and GB200 NVL72 platforms deliver breakthrough AI and HPC performance, thanks to tightly integrated technologies such as NVIDIA NVLink™, ultra-fast networking, and full-stack AI software optimizations. These platforms are engineered to support the most compute-intensive environments—ones that demand multiple GPU nodes working in concert with minimal communication overhead.
Key highlights of the Storage Scale System 6000 include:
- Up to 330 GB/s throughput with ultra-low latency
- Up to 13 million IOPS via NVMe over Fabrics (NVMeoF)
- Up to 1.8 PB effective capacity in a compact 4U form factor
With this latest NVIDIA GB200 NVL72 certification, IBM reinforces its commitment to supporting the next generation of generative AI, simulation, and analytics workloads with unmatched storage performance.
For more information, please see NVIDIA GB200 NVL72 Systems
High-Performance Storage Reference Architecture for IBM Storage Scale System 6000.
For more information on IBM Storage Solutions for NVIDIA AI, See: https://www.ibm.com/solutions/storage/nvidia