ONTAP AI – NVIDIA and NetApp solution

ONTAP AI combines NVIDIA DGX servers, NVIDIA Tesla V100 graphics accelerators, NetApp AFF A800 all-flash storage system along with state-of-the-art Cisco switches. The design of this solution thus greatly simplifies and accelerates the deployment of artificial intelligence (AI).

netapp_owler_20160621_123036_original-1024x307

NVIDIA DGX supercomputers

At the heart of ONTAP AI is the NVIDIA DGX supercomputer, a fully integrated hardware and software system designed for deep learning. The DGX platform uses the NVIDIA GPU Cloud Deep Learning Software Stack, which is optimized for maximum performance using graphics accelerators.

More information

Superfast NetApp AFF storage

NetApp AFF systems are some of the fastest and most flexible all-flash storage in the world and feature NVMe technologies. NetApp AFF A800 storage is capable of sending data to NVIDIA DGX-1 systems at speeds up to 300GBps, which is about four times faster than other competing solutions.

Reliable Cisco networking

Cisco Nexus 3232C 100Gb Ethernet switch is a low latency, dense, high-performance, power-efficient, 100-Gbps switch designed for the data centre.

Scalable architecture

ONTAP AI is a highly flexible solution that can be tailored to the specific needs of the user and can be expanded as those needs grow.

ONTAP-AI DGX-1 reference architecture

NVIDIA DGX-1 servers are an extremely powerful deep learning and AI platform that leverages equally powerful storage and networking infrastructure to deliver the highest possible performance. By combining NetApp AFF systems with Cisco Nexus switches, you can implement this architecture at almost any scale, from a single DGX-1 paired with an AFF A220 to 96 DGX-1 servers on a 12-node NetApp AFF A800 storage cluster.

ONTAP-AI DGX-1 reference architecture

ONTAP-AI DGX-1 reference architecture

Wide configuration options:

NetApp ONTAP AI Powered by NVIDIA - nva-1121-design pdf

ONTAP-AI DGX-2 reference architecture

NVIDIA DGX-2 servers are an extremely powerful deep learning and AI platform that leverages equally powerful storage and networking infrastructure to deliver the highest possible performance. By combining NetApp AFF systems with Cisco Nexus switches, you can implement this architecture at almost any scale, from a single DGX-2 paired with an AFF A220 to 4 DGX-2 servers on a multi-node NetApp AFF A800 storage cluster.

ONTAP AI DGX-2 reference architecture

ONTAP-AI DGX-2 reference architecture

Additional configuration options:

NVIDIA offers special pricing and project-specific programs on both Tesla cards and DGX systems, plus support for educational institutions (EDUs) and start-ups.

Get an offer

Testing

To test the performance and especially the speed of deploying ML and AI applications, we have the NVIDIA DGX Station and, as part of the NVIDIA Test Drive program, 2× NVIDIA Tesla V100. If you are interested in our testing offer, please fill out this form.