NISI Developing Innovative Medical Technologies with NetApp ONTAP AI
Mike McNamara
NISI first used an NVIDIA DGX-1 as a high-performance computing node to help with its survey of research courses. Because the initial training dataset was not large, the simulation was done entirely on the NVIDIA DGX-1. However, NISI was collecting more labeled medical images for training its new network and sourcing data from several hospitals and clinical centers, and its data volumes were growing exponentially.
In addition to the images, NISI also needed to collect videos showing the times when the images were taken. These metrics enhanced the overall system and performance, and also its data storage requirements. NISI required larger storage and more data throughput to facilitate its deep learning (DL) process. Its previous storage network throughput speeds clocked around 1 gigabit per second, which was not nearly fast enough for its DL solution.
NISI is a startup with limited resources. NetApp was able to provide the company a cost-effective but high-tier solution. In conjunction with the NVIDIA DGX-1 already deployed by NISI, NetApp provided ONTAP® AI with a NetApp AFF A300 all-flash array that uses predictive AI. The system improves performance, speed, and IT agility for end users to carry out deep learning and AI workloads. The Mellanox SN2100 100Gb Ethernet switches facilitated the connection between the NetApp AFF A300 and NVIDIA DGX-1 computing node. All the components were optimized by the NetApp ONTAP AI reference architecture, which was pivotal for seamless setup, configuration, and troubleshooting throughout the deployment process
NISI was able to migrate data rapidly from the DGX-1 to the new NetApp storage. These technologies have created seamless and transparent operations for the technology team and AI researchers. Furthermore, NISI now has spare computing performance and storage bandwidth capabilities and is currently migrating some of its business networks to the NetApp solution, such as its enterprise resource planning (ERP) system.
To learn more about the NISI story and analyst IDC’s perspective, read the report.
Mike McNamara
後續步驟
.png?width=5760&format=avif&disable=upscale)