ACCELERATED,
ANY-SCALE AI
Providing unmatched storage performance and flexibility for Artificial Intelligence and Deep Learning.
AI & DEEP LEARNING STORAGE AI & DEEP LEARNING STORAGEA3i: ACCELERATED, ANY-SCALE AI SOLUTIONS FROM DDN
Engineered from the ground up for the AI-enabled data center, DDN’s A3I solutions are fully optimized to handle the spectrum of AI and DL activities concurrently: data ingest and preparation, training, validation, and inference. The DDN A3I platform is easy to deploy and manage, highly scalable in both performance and capacity, and represents a highly efficient and resilient solution for all of your current and future AI requirements.
USE CASES
Unmatched NVIDIA® DGX-1™ Performance
DDN A3I solutions greatly simplify the use of the DGX-1 server, while also delivering performance and efficiency for full GPU saturation, and high levels of scalability.
Fully Integrated, GPU-Optimized Data Platform
Easy to deploy, DDN A3I solutions are turn-key, pre-configured, and provide the most capable scale-out platform for capacity and performance.
Full GPU Saturation
Fully optimized for all types of I/O patterns and data layouts, A3I solutions deliver data to applications, ensuring full GPU resource utilization even with distributed applications running on multiple computing servers.
Capacity-Efficient
AI Storage
A3I solutions provide up to 360TB of reliable and efficient dual ported NVMe flash in a high-density 2U configuration and can scale seamlessly to multiple petabytes of flash.
Highest Resiliency, Reliability, Security at Scale
Engineered to provide the highest data availability and maximum system uptime, all A3I hardware and software components are integrated as a fully redundant system.
Unified Namespace
A3I solutions allow for consolidation of hot training data and warm expanding data libraries into a single platform, providing easy data access from a unified interface
Multi-Tenancy and Quotas
A3I solutions can be secured on a per-tenant basis that ensure users and applications can only access the data that they’re entitled to. Advanced quota controls provide easy management of file system consumption at the user, group, and project level.
WHITEPAPER: DDN ACCELERATION FOR ARTIFICIAL INTELLIGENCE AND DEEP LEARNING APPLICATIONS
With DDN A3I solutions, data scientists, researchers and AI team members are able to focus their full attention on what really matters – turning data into valuable insights at unprecedented velocity and accuracy.
DDN A3I solutions enable and accelerate end-to-end data pipelines for AI and DL workflows of all scale running on fully integrated GPU compute platforms. They are designed to provide extreme amounts of performance and capacity through a tight integration between DDN and NVIDIA platforms. Every layer of hardware and software engaged in delivering and storing data is optimized for fast, responsive and reliable file access.
DDN AI200 and AI400: Compact, all-flash appliances optimized for intensive I/O workloads.
Up to 360TB of dual port NVMe flash in 2U
Up to 20GB/s throughput, up to 1 million IOPS (AI200)
Up to 40GB/s throughput, up to 3 million IOPS (AI400)
Ideal for flash-only highest possible IOPS and GB/s, high-performance multi-GPU computing, and easy deployment and management.
The AI200 and AI400 are the world’s most efficient, reliable and easy to use data storage system for AI and DL applications. They tightly integrate with hard disk tiers to help manage economics as data volumes expand. It is also specifically optimized to keep GPU computing resources fully saturated, ensuring maximum efficiency while easily managing tough data operations, from bursty-ingest to large data transformations.
Extensive testing with widely-used AI and DL applications demonstrates that a single AI200 provides tremendous acceleration for data preparation, neural network training and inference tasks using DGX-1 servers. With AI200, Caffe applications running on a NVIDIA® DGX-1TM server demonstrates 2.4X increased image throughput and twice shorter completion times. With AI200, TensorFlow training applications demonstrate double image throughput and complete twice as fast on a DGX-1 server using a AI200 solution. Testing also demonstrates that the DDN parallel architecture maintains linear performance for applications that leverage distributed computing on multiple DGX-1 servers, such as Horovod.
DDN AI7990: Hybrid flash appliance with unified high-performance and large capacity storage tiers.
20GB/s throughput, 1 million IOPS
Up to 600TB of SSD in 4U
Expand up to 5.4PB in 20U
Ideal for flash and disk hybrid flexibility and scale, high-performance multi-GPU computing, unified workspace and data collection, and scalability of performance and capacity.
DDN’s AI7990 is the world’s most capable hybrid flash and hard disk data storage platform for AI and DL. With true end-to-end parallel access to flash and deeply expandable HDD storage, the AI7990 outperforms NAS platforms and delivers the economics of HDD for your growing data library. The AI7990 keeps DGX-1 servers saturated with data, ensuring maximum utilization while managing tough data operations from bursty-ingest to large-scale data transformations.
Unleash the Power of Your AI
DDN has equipped its laboratories with leading GPU compute platforms to provide unique benchmarking and testing capabilities for AI and DL applications. Our team of DDN experts also perform an in-depth evaluation of requirements and provide application-level optimization of data workflows for your project. They will then design and propose an optimized, highly reliable and easy to use solution that best enables and accelerates the customer effort.
Contact DDN today and engage our team of experts to unleash the power of your AI projects.