An AI factory without data is like an engine without fuel; it simply can’t run. IBM Storage Scale System 6000 solves this by unifying data across edge, core, and cloud through its global namespace and Active File Management (AFM) capabilities. By eliminating silos and accelerating access, it enables AI pipelines to start immediately and deliver results faster.

In today’s modern data center, unlocking “dark data” and eliminating performance bottlenecks for AI inferencing are top priorities for IT leaders. To address both concerns for our clients, today we’re introducing an updated version of IBM Storage Scale System 6000, IBM’s high-performance elastic data system, and a new expansion enclosure that significantly increases total capacity. These enhancements will help organizations deploy data-hungry AI workloads, and when combined with the NVIDIA BlueField data processing unit (DPU), run their most performance-intensive workloads at scale.​

Tripling Maximum Capacity for the IBM Storage Scale System 6000

With the updated Storage Scale System 6000, IBM has tripled the maximum capacity to 47PB per rack by adding support for industry-standard QLC flash storage – in 30TB, 60TB and 122TB SSD configurations, providing clients more data storage options to fit their needs. With the introduction of IBM Storage Scale System 7.0.0 software, organizations can leverage the following new updates:

• Optimize data management with multi-flash tiering – adding support for industry-standard NVMe TLC and QLC flash drives in the same system.
• Feed applications faster with a data acceleration tier that leverages high-performance NVMe-oF to deliver up to 340 gigabytes per second throughput and up to 28M IOPS. It is engineered to offer exceptional compression capability and data density while drawing less power.
• Enhanced data protection with an enhanced framework that comes with wider erasure code support (16+2/3P) for increased disk efficiency and performance and increased write performance.​
• Train and optimize AI faster with NVIDIA Spectrum-X Ethernet networking that can significantly reduce the time required for checkpoints when training AI foundation models. ​

Bringing an All-Flash Expansion Enclosure to IBM Scale System

To address workloads that operate on massive data sets, we’re introducing the IBM Scale System All-Flash Expansion Enclosure, which is optimized for high performance AI training, data inferencing, HPC, and data intensive workloads. With the introduction of 122TB QLC NVMe SSDs, the enhanced Storage Scale System 6000 delivers more than 47PB of cost-effective high-density flash capacity in a single 42U rack with the All-Flash Expansion Enclosure.

The enclosure also integrates with NVIDIA BlueField-3 DPUs for maximum performance at scale. Details include:

• Adapters enable up to 400Gb/s of high-performance network throughput and NVMe-oF offload capability for the Storage Scale System.
• Each expansion enclosure can deliver up to 100GB/s of throughput and more than 3PB of raw flash capacity in a 2U form factor.
• An enclosure can support up to four NVIDIA BlueField-3 DPUs and 26 dual-port industry standard QLC flash drives in a 2U form factor. 

The scalability of the new All-Flash Expansion enclosure also supports bigger caches and multitenant service provider requirements. With IBM Storage Scale System 6000, multitenancy is highly configurable at the cluster, file system, or fileset level. This gives service providers and large-scale AI factories the ability to isolate workloads, enforce security policies, and optimize resource allocation without sacrificing performance. Whether supporting supercomputers, HPC environments, or AI-as-a-service offerings, this architecture ensures secure, scalable, and cost-efficient multitenant operations across diverse data-intensive workloads.

IT infrastructure teams are faced with unprecedented data challenges to scale AI while continuing to support all existing workloads. The enhanced IBM Storage Scale System 6000 addresses these challenges and enables globally dispersed teams to collaborate on data regardless of protocol, location, or format. ​Tripling capacity and adding the new expansion enclosure allow AFM, our global-caching and data‑orchestration layer, to cache much larger working sets of data closer to GPUs, eliminating data silos and enabling AI pipelines to run faster.

The IBM Storage Scale System 7.0.0 software upgrade will be available on December 9^th, and general availability for the new IBM Storage Scale System All-Flash Expansion Enclosure is December 12^th. To learn more, visit https://www.ibm.com/products/storage-scale-system, or visit our booth (#917) at SC25.