VAST Data Achieves NVIDIA DGX Sc

• 18 May 2023
• TM Team
• 0 Comments

VAST Data  announced that VAST’s Data Platform is now a certified datastore for NVIDIA DGX SuperPOD. This new validation provides organizations with agile, scalable performance for the most challenging machine learning and deep learning workloads, including generative AI.

The VAST Data Platform brings to market the first enterprise network attached storage (NAS) solution approved to support NVIDIA DGX SuperPOD. NVIDIA customers can now enjoy the simplicity of the VAST data platform with virtually limitless levels of scale and performance via a system architecture that eliminates data tiers and infrastructure silos to make large scale AI simpler, faster and easier to manage.

With VAST, customers go beyond the definitions of classical high performance computing (HPC) storage with an exabyte-scale data platform that provides enterprise-levels of systems and data management, featuring capabilities such as:

·         Encryption, authentication and external key management along with a data catalog to help customers manage their data and data security in real time.

·         VAST’s multi-tenant approach is ideal for GPU service providers by providing tenant domains for securing and isolating data as well as tenant-level QOS controls to provide performance fairness.

·         VAST systems save customers money by being designed for hyperscale QLC flash and by applying next-generation global data reduction algorithms, customers on average realize data reduction of nearly 3:1 across VAST’s fleet of deployments.

·         VAST systems better protect data by providing support for n-1 and 1-n replication topologies and up to 1 million ransomware-proof snapshots.

·         System expansions are always online, as are software upgrades.

This feature-rich platform delivers tremendous value to customers, so much so that VAST has received a “100 percent willingness to recommend” by its customers, as reported by Gartner Peer Insights.

This announcement regarding NVIDIA DGX SuperPOD is the culmination of a long-standing collaboration between VAST Data and NVIDIA that dates back to 2016, when the two companies began work on building disaggregated systems and delivering innovative AI solutions to customers. This contributed to the invention of VAST’s disaggregated, shared-everything (DASE) architecture, which uniquely positioned VAST as a provider of parallel and resilient infrastructure that makes standard NAS protocols suitable for DGX-class computing. Since its announcement, VAST’s DASE architecture has been described by IDC as “the storage architecture of the future.” The collaboration then deepened as VAST developed support for NVIDIA GPUDirect Storage and announced support for NVIDIA BlueField data processing units to enable customers to step easily into low-cost and modular flash infrastructure.

“VAST’s alliance and growing momentum with NVIDIA to help customers solve their greatest AI challenges takes another big step forward today,” said Renen Hallak, Co-Founder and CEO of VAST Data. “As AI and HPC interest, adoption, and implementation has exploded over the last year, the need for accelerated infrastructure has never been more acute. The VAST data platform brings to market a turnkey AI data center solution that is enabling the future of AI.”

“As enterprise AI and HPC adoption grows, organizations are seeking full-stack infrastructure that delivers performance and scale to turn their data into insight,” said Charlie Boyle, Vice President of DGX systems at NVIDIA. “Our collaboration with VAST Data enables customers to use the VAST Data Platform with their NVIDIA DGX SuperPOD environments to accelerate modern enterprise and cloud-based AI initiatives.”

“VAST and NVIDIA represent a formidable partnership in powering the next generation of AI and ML applications,” said Peter Rutten, Performance-Intensive Computing Solutions Global Research Lead at IDC. “Together they deliver converged solutions that flexibly offer the capacity and exabyte scale customers require while simplifying AI infrastructure and data management to allow data scientists to turn information into insight quickly.”