IBM releases Elastic Storage System 3500 for AI Workloads

Global technology giant IBM announces the launch of its new Elastic storage system 3500, which is specifically built to handle high-performance artificial intelligence (AI) and modern data analytics workloads. 

The newly launched storage system outperforms the rest of the company’s software-defined portfolio thanks to a mix of new hardware and the IBM Spectrum Scale. 

IBM Elastic Storage System (ESS) 3500 combines IBM Spectrum Scale for ESS with high-performance Non-Volatile Memory Express (NVMe) Generation 4 storage in a 2U form factor. 

Read More: Diane Staheli joins DoD as new Chief of Responsible AI

According to the company, its new solution improves the speed and availability of other workloads while also accelerating AI workloads and optimizing GPU compute settings. 

“ESS 3500 is built on a flash-optimized design with an end-to-end strategy to bring extremely high throughput and low latency to AI, machine learning, deep learning, and multi-cloud environments. ESS 3500 is designed from the ground up with AI in mind, combining innovative, flexible, and powerful NVMe storage built with IBM Spectrum Scale to meet the challenges of keeping your GPU-rich compute servers fully utilized,” mentioned IBM. 

Moreover, other workloads and computing environments benefit from the ESS 3500’s industry-leading performance and dependability. According to the company’s spec sheet, the flash storage uses PCIe Gen 4 SSDs with a throughput of up to 91 Gbps when combined with the CPU. 

The array can expand from 46 TB to over 1 PB utilizing a 2.5 compression rate, and the sustained throughput can be raised to 1.8 TBps when scaling the ESS 3500 to a cluster of 20 units.