Dell Technologies Advances AI and Generative AI Strategies with Accelerated Storage Innovations
-
Dell Technologies recently announced new enterprise data storage technologies validated with NVIDIA DGX SuperPod AI infrastructure to help customers achieve faster AI and generative AI performance. Arthur Lewis, President of Dell Technologies Infrastructure Solutions Group, stated: "Storage performance is critical for successful AI and generative AI outcomes. Customers rely on us to continuously push the boundaries of storage innovation, including eliminating data access bottlenecks that limit throughput and scalability for compute-intensive applications. We address these needs by delivering fast, efficient, and secure access to data – transforming it into a goldmine of AI and generative AI possibilities."
To meet the high-performance and efficiency demands of AI storage, Dell PowerScale – the world's most flexible, secure, and efficient distributed file storage system – has introduced new advancements. With PowerScale OneFS software enhancements, enterprises can now prepare, train, optimize, and infer AI models faster. The new all-flash PowerScale storage systems powered by Dell's latest generation PowerEdge servers deliver up to 2x performance improvement in streaming read/write operations for customers.
PowerScale will also introduce new intelligent scaling capabilities that boost performance per compute node and enhance GPU utilization, enabling faster storage throughput for AI training, checkpointing, and inference tasks.
Through its collaboration with NVIDIA, Dell PowerScale is being validated with NVIDIA DGX SuperPOD, enabling customers to leverage proven NVIDIA DGX systems, Dell PowerScale storage, and NVIDIA Quantum-2 InfiniBand as well as Spectrum Ethernet networks for faster and more efficient AI storage. Dell's solution is expected to become the first Ethernet storage solution validated on NVIDIA DGX SuperPOD.
Furthermore, Dell is committed to bringing AI wherever data resides. With nearly 87% of companies adopting a multi-cloud strategy, Dell provides customers the freedom to process data on-premises, at the edge, or in the public cloud. The launch of Dell Apex File Storage for Microsoft Azure is the latest addition to Dell Apex public cloud storage portfolio, offering enterprise-grade file performance and management capabilities within Microsoft Azure. Customers can easily meet the demands of performance-intensive AI and machine learning applications such as Azure OpenAI Service and Azure AI Vision.
Dell Apex File Storage in AWS and Azure allows customers to leverage cloud-native AI and generative AI workflows, improving data access and movement for public cloud or on-premises data. Through its partnership with Databricks, customers can choose from a variety of large language models (LLMs) on proprietary data stored in Dell Apex File Storage and retrain foundational models using the Databricks MosaicML library, providing flexibility for multi-cloud environments.
This partnership with Databricks represents an extension of Dell's data management ecosystem collaborations, aimed at helping customers accelerate AI and analytics workloads. Additionally, through Dell's planned open, modern data lake solutions, customers will be able to utilize Starburst analytics software integrated with Dell PowerEdge servers, Dell PowerScale, Dell ECS, and Dell ObjectScale file and object storage systems to extract insights from stored data.
The aforementioned Dell PowerScale OneFS software enhancements are scheduled for global release this month, while the next-generation all-flash Dell PowerScale systems and new intelligent scaling features are planned for global availability in the first half of 2024. Dell PowerScale is currently being validated with Nvidia DGX SuperPOD, with completion expected in Q1 2024. Dell Apex for Microsoft Azure file storage is planned for global release in the first half of 2024. Dell Apex file storage integration for AWS and Azure is available today on AWS, with Microsoft Azure integration planned for global release in the first half of 2024.