IBM Storage Scale System 6000 Adds Certification for NVIDIA GB200 NVL72 Platform

By Ted Hoover, product manager, storage for data, AI, HPC, IBM Corp.

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.

Storage Scale System 6000 front

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 Storage Scale System 6000 include:

• Up to 330GB/s throughput with ultra-low latency
• Up to 13 million IOs via NVMe over Fabrics (NVMeoF)
• Up to 1.8PB effective capacity in a compact 4U form factor

With this latest NVIDIA GB200 NVL72 certification, IBM Corp. reinforces its commitment to supporting the next-gen of GenAI, simulation, and analytics workloads with unmatched storage performance.

Resources:
NVIDIA GB200 NVL72 Systems   
High-Performance Storage Reference Architecture for IBM Storage Scale System 6000     
IBM Storage Solutions for NVIDIA AI