DDN in AI Initiatives With NVIDIA DGX Systems

Enterprises increasingly rely on AI as a powerful business enabler with massive potential rewards.

However, research has shown that a staggering nine out of 10 AI initiatives fail due to complexities in architecting and optimizing AI deployments, and failing to successfully take them from prototype to production. ⁽¹⁾

DDN (DataDirect Networks, Inc.) takes the integration of AI in Enterprise IT infrastructures to the next level, by delivering easy to deploy production-ready and cost-effective AI storage systems to customers. The company‘s A³I (Accelerated, Any-Scale AI) solutions bring performance and control to AI workflows at any scale.

“To achieve a strong competitive edge, enterprises must extract more value out of their data and greatly accelerate time to insight. For that to happen, a flexible and agile storage platform which reliably and intelligently handles massive amounts of dynamic data is essential,” said Dr. James Coomer, VP, products. “For the past 2 decades DDN has successfully deployed value driven intelligent infrastructure storage solutions. Our increasing success in bringing AI-enabling storage to more than 11,000 customers reinforces our reputation as a leader in At-Scale solutions.“

DDN and Nvidia Corp. offer a joint integrated AI solution that is optimized across storage, software, networking and GPU acceleration. As the only storage platform certified by Nvidia for DGX SuperPOD with DGX A100 systems, the firm’s A³I handles hundreds of petabytes of dynamic data and is optimized for the end-to-end AI workload.

In addition to DGX SuperPOD support, the company’s A³I portfolio has grown to include DGX POD offering entry-level infrastructure solutions for enterprise, starting with 2-node DGX A100 configurations for businesses that want a starter system that offers modular scale as AI workloads grow.

Joint solution delivers value to customers through:

• 33x faster AI inference than traditional storage

• 10x faster AI data ingestion

• 46x more IO/s into GPU Infrastructure ⁽²⁾

“Advanced AI and data science applications process massive amounts of data, making it critical for storage to keep up with the accelerated computing of Nvidia DGX systems,” said Tony Paikeday, senior director, DGX systems, Nvidia. “The DDN A³I reference architectures for NVIDIA DGX SuperPOD and DGX POD enables organizations to rapidly implement the high-performance AI infrastructure they need to power their most important work at any level of scale.“

The A³I systems combine the company’s storage and AI-powered data management tools to deliver acceleration in complex AI applications such as recommendation systems, image analysis and natural language processing (NLP). It is optimized to accelerate ML, streamline deep learning workflows and eliminate complexity in deployment and management at any scale.

“AI and ML-driven big data analytics are a popular next gen application being deployed by enterprises undergoing digital transformation,” said Eric Burgener, research VP, infrastructure systems, platforms and technologies, IDC. “The introduction of this new, smaller NVIDIA DGX POD reference architecture makes it much easier to deploy the infrastructure to get AI workloads up and running quickly so they can start driving business value.“

A³I client plug-ins for DGX systems and other GPU environments ensure applications run fast out of the box with AI container workloads. By delivering an optimized data path, A³I solutions overcome I/O difficulties previously considered insurmountable, accelerating performance intensive applications and scalable data science use cases.

“Undeniably, since AI has entered our lives, it has greatly facilitated our social development and research capabilities, and has facilitated breakthroughs across all industries,” said Dr. Akkarit Sangpetch, ECE program director, CMKL University, and member, Faculty of Engineering, KMIT. “AI, furthermore, has proved its great ability to identify patterns in the data, and we are committed to offering the fastest and most sophisticated infrastructure to unlock insights contained in those massive datasets. DDN’s solution gives us the ability to do just that. In addition, their trusted reputation and deep experience with educational institutes further affirm our goal in providing an AI data platform for research and education.“

Resources:
A³I solutions
A³I/NVIDIA DGX POD Reference Architecture

⁽¹⁾  Deloitte 2020 Survey: State of AI in the Enterprise; Third Edition: Thriving in the Era of Pervasive AI
⁽²⁾  Results based on DDN internal testing; see Removing Risk from AI Scale, DDN blog