SC22:  DDN Unveils Reference Architecture in Collaboration with Atos for BullSequana XH3000 Supercomputer

A the SC22, DDN (DataDirect Networks, Inc.) announced a Reference Architecture in collaboration with Atos SE.

The company also introduced enhancements to its scalable and flexible monitoring interface, DDN Insight, to streamline and simplify management and support of HPC and AI infrastructures at-scale.

The firm worked in tight collaboration with Atos on this Reference Architecture to advance company’s goal of simplifying the adoption of advanced AI infrastructure for enterprises and research institutions. Historically, storage systems were often chosen to complement compute infrastructure as it was being deployed. This approach leads to silos of data and requires data movement to use the optimal compute platform for a certain stage of processing. The firm’s approach allows data to stay in a single manageable system, regardless of the compute technology used, with scalable performance and capacity to accommodate any rate of growth.

The collaboration between the two companies is founded on a long relationship. Having deployed some of the world’s largest supercomputing systems together, the integration of A³I AI400X2 appliances with Atos’s BullSequana XH3000 systems is a natural continuation of that relationship. With the firm‘s parallel file system delivering data with high-throughput, low-latency and massive concurrency for processing by the XH3000 systems, customers can be assured that they are maximizing the performance of their applications. Combining GPU processing, HPC software suites, Atos next-generation (DLC) Direct Liquid Cooling technology, and an architecture that is flexible, dense and secured by design, BullSequana XH3000 delivers both performance and efficiency, which is complemented by the company’s small footprint and economical performance.

“Atos is pleased to pursue and strengthen our current strategic partnership with DDN data storage solutions, which are a key element for the high-performance AI computing systems we are providing to our customers,” said Dr Cédric Bourrasset, head, high-performance AI computing, Atos. “This new Reference Architecture, based on our new BullSequana XH3000 and DDN A³I Appliances, will bring the power efficiency required to address the future training challenges with the next generation of AI models.“

The company’s Insight helps organizations see across their infrastructure and workloads to improve service levels and find the root cause of complex issues faster. Insight can monitor hardware on workloads down to the individual job level from a single dashboard. The 4.2 release delivers better visualization of components and real-time operation of an EXAScaler environment.

“DDN systems address some of the largest data challenges on earth,” said Dr. James Coomer, SVP product, DDN. “Our goal is to simplify the storage and management associated with large AI, HPC and analytics data sets. These Reference Architectures and the enhancements to DDN Insight allow customers to quickly adopt and put into production new technologies, streamline their infrastructure management and take proactive steps to keep their environment healthy.“

Resource:
Atos BullSequana XH3000 and DDN A³I Reference Architecture