ISC23: DDN QLC SSD Storage Delivers 10x Speed for AI and Data Centers at Any Scale

DDN (DataDirect Networks Inc.) announced a  breakthrough for its all flash and hybrid storage solutions.

The company’s parallel file system technology, combined with AI and data center specific data compression, delivers high performance efficiency straight into generative AI, ML and other enterprise high-performance applications.

Eliminating the need for complex networking and heavily bottlenecked performance found in other data storage solutions, the company’s AI400X2 QLC and hybrid storage arrays combine the firm’s parallel file system with novel client-side data compression, increasing performance by 10x, growing effective capacity by up to 15x and reducing data center footprint by 2x.

By owning and optimizing the entire data path, the company’s appliances directly impact application performance. In addition, the platform requires less infrastructure, reduces power draw and lowers the consumption of data center real estate.

“Today’s QLC scale-out NAS systems offer low cost and high capacity, but they are extremely inefficient with IOs, throughput and latencies, making them unusable for high-performance environments such as AI, machine learning, and real-time applications,” said Dr. James Coomer, SVP, products. “Our parallel file system and data compression technologies which power DDN’s new AI400X2 QLC and hybrid storage arrays, solve the challenges in these at-scale and high-performance environments, delivering a magnitude of improvements and benefits for our customers.”

AI storage requires purpose-built infrastructure designed to accommodate data-hungry workflows. While generative AI is grabbing headlines, other promising applications like ultra-realistic 3D and immersive universes in gaming, sophisticated new protein and molecule creations for drug discovery, and autonomous driving are also driving the need for efficient, dependable performance that can readily scale.

Starting as a single 2U appliance with no additional components, customers can expand storage capabilities by adding additional appliances or enclosures without rearchitecting to meet expanding performance or capacity requirements. Furthermore, the firm’s EXAScaler parallel file system with client-side compression can improve performance by trading network traffic for more available CPU cycles. The company’s QLC storage platform is a solution to make the management of generative AI achievable.

As a provider in AI and multi-cloud data management solutions, the company offers high performance and flexible storage, optimized storage for any workflow at any scale. The firm’s all-TLC flash-based systems deliver the best IO/s and throughput per rack, with up to 70 million IO/s in a single rack. Its hybrid solution with TLC flash and traditional disk drives offers excellent price/TB at 20PB/rack. The QLC flash system provides a best price/flash/TB with up to 14 million IO/s in a single rack and the highest storage density.

The AI400X2 QLC systems range from 2.9PB to 7.3TB and effective capacity from 4.7PB to 11.7PB using a conservative compression rate of 2x. The company intends to begin shipping its QLC-based solutions in 3Q23.

DDN at ISC23
The company is exhibiting at ISC High Performance 2023 from May 22 to 25 in Hamburg, Germany, and will offer technical presentations and panels and offer visitors face-to-face time with experts. 

Resources:
DDN EXAScaler     
DDN Storage Fusion Architecture (SFA)         
DDN at ISC23