AI Datasheet – faster deep learning for artificial intelligence and analytics

Artificial Intelligence (AI) data pipelines are inherently different from traditional file-based IO applications. Each stage within AI data pipelines has distinct storage IO requirements: massive bandwidth for ingest and training; mixed read/write handling for extract, transform, load (ETL); ultra-low latency for inference; and a single namespace for entire data pipeline visibility. Furthermore, AI at the edge is driving the need for edge-to-core-to-cloud data pipelines. Hence, an ideal data storage solution must meet all these varied requirements and deliver timely insights at scale. Traditional solutions lack these capabilities and often fall short in meeting performance, data shareability, and mobility requirements. DataOps is data management for the AI era and delivers operational agility, governance, and actionable intelligence for these data pipelines by breaking silos.