AI is transforming workflows across industries, accelerating research, optimizing manufacturing, and creating new products in financial services.  While AI is opening opportunities for almost every business, it is also a threat to those organizations that don’t transform quickly enough.  Repeatedly cited as a major barrier to the success of AI projects, infrastructure that can grow with escalating requirements is a must for data-driven organizations.

Deploying systems that deliver predictable, reliable performance to accelerate data science workflows does not need to be daunting or time consuming.  NVIDIA DGX SuperPOD, integrating DDN’s A3I storage solutions, is the fastest path to scalable enterprise AI infrastructure.  Designed and deployed in production to solve some of the most challenging data science challenges, the combination of GPU-accelerated computing with NVIDIA DGX A100 systems and the shared parallel storage architecture of the DDN AI400X is now available to all organizations as rapidly deployable supercomputing infrastructure, backed by the expertise of the leaders in AI computing and the leaders in AI storage.  A complete reference architecture removes much of the complexity associated with the design, procurement and deployment of systems at this scale.

One key to unlocking the power of all the GPU computing resources available with DGX SuperPOD is delivering consistent performance through the use of a massively parallel file system.  AI application performance can be highly limited by how fast data can be read and re-read from storage.  Typical enterprise storage systems rely on NFS to deliver data, a protocol that encounters performance limits very quickly and does not scale simply or efficiently.  DDN and NVIDIA have worked together to deliver a highly optimized data path that ensures all GPUs in the computing cluster are used effectively at scale. Validated by extensive testing and in-production deployment with NVIDIA Selene, the company’s own DGX SuperPOD deployment, DDN A3I is the scalable AI accelerator.

Beyond performance, DDN’s A3I solutions meet the needs of data-intensive enterprises by satisfying all the requirements for a centralized AI data repository.  Data movement is inefficient and storage silos can limit the ability of new applications to integrate disparate data to gain new insight.  DDN has designed A3I systems to address all stages of the AI workflow, from data acquisition to cleaning and cataloging, processing, analysis and archive. Combining the power of the parallel files system with standard storage protocols like NFS and SMB, object storage, and multicloud deployment, A3I allows for continuous deep learning at scale.  Moving from capture to analysis on a single platform greatly accelerates time to insight.

DDN’s established position at the top of the AI and HPC storage market comes from a long-term investment in leading-edge research and development that solves our customers’ end-to-end data needs. We are continuously driving to deliver further efficiency and visibility into our customers’ environments.  Another example of this is DDN’s Insight, which provides full visibility into the entire data path.

Even the most optimized applications can encounter bottlenecks.  Insight allows customers to quickly identify and diagnose jobs that are underperforming or over consuming resources – no matter how many concurrent jobs are running.  As with everything DDN engineers, Insight is ready to support the largest AI environments imaginable.

With over 20 years of experience and expertise delivering solutions to the most demanding organizations worldwide, DDN is the perfect complement to the capabilities of NVIDIA’s DGX A100 systems, especially when deployed at the scale of DGX SuperPOD scale.  For the optimal AI infrastructure experience, contact DDN today.

To learn more, join DDN at GTC 2020 in our session, “Maximizing AI Success with Accelerated Data,” taking place Tuesday, October 6 at 1 pm PT and Thursday, October 8 at 9 am PT.

  • Kurt Kuckein
  • Kurt Kuckein
  • Vice President, Marketing
  • Date: October 5, 2020

This site uses cookies. Please see our Privacy Policy to learn more about how we use cookies and how to change your settings if you do not want cookies on your computer. By continuing to use this site without changing your cookies settings, you consent to the use of our cookies. Privacy Policy