IBM Storage Scale System

IBM Storage Scale System is a high-performance hardware platform running IBM Storage Scale software, designed to help organizations build a global data platform that efficiently handles file and object data for the most demanding artificial intelligence (AI), high-performance computing (HPC), analytics and hybrid cloud workloads.

By combining the power of IBM Storage Scale software with NVMe flash technology, it delivers a fast and flexible solution for managing vast amounts of unstructured data, including documents, audio, images, videos and other objects.

Storage Scale System is built on a massively parallel file system and can be deployed across multiple hardware platforms, including x86, IBM Power, IBM Z®, ARM-based POSIX clients, virtual machines and Kubernetes. It delivers exceptional performance for GPU infrastructure with support for NVIDIA GPUDirect Storage.

Very little enterprise data is used to train the large language models underlying their chatbots and other AI assistants, limiting their business value. The content-aware storage capabilities in IBM Storage Scale address this challenge by extracting the semantic meaning hidden inside unstructured data so that AI assistants can automatically generate smarter answers.

Storage Scale System 6000 is designed for extreme performance. It features:

• a single 4U node with active-active controllers and redundant hardware to maximize uptime;
• up to 330 gigabytes per second (GB/S) throughput with low latency;
• up to 13 million IOPS with the NVMeoF performance tier; and
  
• up to 3.4PBe (effective capacity) in a standard 4U rack space.

Storage Scale System 6000 is engineered to scale to thousands of nodes and petabytes of capacity. It runs IBM Storage Scale RAID erasure coding, which provides data efficiency, consistent high performance, mitigation of storage hardware failures, intelligent monitoring and dynamic tuning of IBM Storage Scale System and IBM Storage Scale data. Installations and updates are delivered by containerized software that speeds and simplifies the maintenance process.

To accelerate the processing of globally distributed data, Storage Scale System 6000 supports the NVIDIA GPUdirect Storage protocol, which enables a direct data path between GPU memory and local or remote storage, such as NVMe or NVMe over Fabric (NVMe-oF).

Storage Scale System 6000 uses a simple building-block approach with performance that scales linearly. That is, a cluster of 10 Storage Scale System 6000 systems is capable of more than 3 terabytes per second throughput. It supports up to nine SAS hard disk drive expansion enclosures.

IBM Storage Scale System 3500 is for customers who require an entry-level or midlevel system for less extreme product environments. It has many of the same capabilities as the Storage Scale System 6000, with differences mainly in size and capacity.