Fueled by data, infrastructure advances, and the ubiquity of machine learning and deep learning (ML/DL) toolkits, artificial Intelligence (AI) solutions are fast becoming a mainstay in the enterprise data center. AI turns data into insights across a broad swath of enterprise verticals as diverse as automotive, healthcare, life sciences, finances, technology, retail, and beyond. Data is now a competitive advantage in industries such as insurance − where predictive AI removes risks from underwriting, finance − where real-time deep-learning recognizes fraud as it happens, and even data center management − where patterns are analyzed to predict failures and scalability issues.
Artificial Intelligence and especially deep learning bring new demands to how data is served to the compute engines that consume it. The new realities of deploying artificial intelligence in the data center change the demands of density, throughput, concurrency and even scale-out data architecture change. IT must think differently about marrying storage and compute to deliver on the promise of AI for the enterprise.
This paper describes how deep learning and artificial intelligence in the enterprise bring new workflows and challenges to data center architecture. It also addresses how solutions can be constructed from infrastructure architectures specifically designed to bring scale-out compute and storage closer together.
Deep learning requires large amounts of data to be fed into the processor without making the processors wait for that data. Properly marrying compute with the right storage technology, such as the Dell EMC Isilon series, allows data to be fed into the machine learning pipeline at the speed of the processor. Properly balanced systems accelerate innovation and deliver flexibility and agility to both IT organizations and the data scientists who rely on them.