Home Enterprise VAST Data Partners with Supermicro, Introduces AI Cloud Architecture Using NVIDIA BlueField-3 DPU Technology

VAST Data Partners with Supermicro, Introduces AI Cloud Architecture Using NVIDIA BlueField-3 DPU Technology

by Harold Fritts

VAST Data has partnered with Supermicro to provide an AI solution designed for simplicity and scale, utilizing NVIDIA-certified systems. This collaboration offers service providers, tech giants, and data-driven enterprises a robust and efficient AI system.

VAST Data has partnered with Supermicro to provide an AI solution designed for simplicity and scale, utilizing NVIDIA-certified systems. This collaboration offers service providers, tech giants, and data-driven enterprises a robust and efficient AI system.

The joint solution promotes ease of use and growth in AI ventures, emphasizing multi-tenancy, impenetrable security, and scalability. The VAST Data Platform enables seamless storage and contextualization of unstructured data and expedites data analysis by combining transactional and analytical warehousing, all while maintaining consistent performance and capacity.

Renen Hallak, the CEO of VAST Data, stated that competitive organizations need partners with a track record in hyperscale. By working with Supermicro, they aim to provide a scalable data platform conducive to the demands of AI computing, utilizing a parallel architecture that supports unstructured and structured data management.

The solution by VAST Data and Supermicro includes:

  • Scale and reliability, with the ability to extend clusters to exabyte levels without compromising uptime.
  • Supports extensive GPU cluster feeding
  • Support for data preparation and training, merging file and object storage with a high-performance database system
  • Global namespace that eliminates data silos and allows for immediate data access and consistency across various locations, enhanced by new edge caching techniques and decentralized lock management.

Charles Liang, CEO of Supermicro, expressed excitement about deploying the VAST Data Platform, which promises substantial deep learning performance when paired with Supermicro’s NVIDIA-Certified Systems.

Supermicro’s systems are well-suited for enterprises, deliver NVIDIA AI Enterprise software support, and offer significant performance boosts. The VAST Data Platform is an all-encompassing software framework for data-centric computing.

VAST Data Uses NVIDIA BlueField-3 DPU Technology for AI Cloud Architecture

VAST Data has also introduced a new AI cloud architecture to enhance performance, service quality, security, and efficiency. It utilizes NVIDIA BlueField-3 DPU technology, turning supercomputers into AI data engines. Collaborating with NVIDIA, VAST Data successfully ported the VAST Data operating system to run natively on BlueField DPUs deployed inside compute servers (e.g., NVIDIA GPU servers). VAST is the first data infrastructure solution that leverages BlueField technology end-to-end.

NVIDIA’s BlueField networking platform merges computing power with hardware accelerators, creating a secure, software-defined accelerated computing infrastructure. This design integrates storage and database processing services into AI servers, enabling scaling to extensive GPU arrays while cutting costs, space, and power demands.

Partnering with NVIDIA, VAST Data’s architecture significantly reduces compute and networking resource needs, saving energy and reducing the data center footprint. It also delivers a parallel storage and database environment to eliminate data service contention, which is crucial for multi-tenant service providers.

Moreover, it reinforces zero-trust security by isolating data management and using standard client protocols. Using NVIDIA DOCA software enables VAST systems to provide block storage services to host operating systems, complementing its file, object, and database services for high-performance applications.

VAST’s co-founder, Jeff Denworth, expressed pride in their collaboration with NVIDIA, highlighting how the architecture embodies their vision for disaggregated data centers. The architecture is being implemented first at CoreWeave, a specialized GPU cloud provider, aiming to scale AI infrastructure for industry leaders.

NVIDIA’s Vice President of Storage Technology commented on the benefits of pairing accelerated computing solutions with accelerated network infrastructure. Meanwhile, CoreWeave’s Vice President of Engineering emphasized the advantages of integrating VAST’s software into their GPU clusters, noting how it streamlines infrastructure and enhances customer experience by optimizing the AI data pipeline. CoreWeave is working to define the future of cloud data management with these advancements.

Engage with StorageReview 

Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | TikTok | RSS Feed