NetApp's Novus hits 100TB/s to ease AI metadata bottlenecks

NetApp's push into production AI centers on a bottleneck that speed alone cannot clear. Syam Nair, the company's chief product officer, told theCUBE at NetApp INSIGHT that while lots of investment is happening and data applications and models are ready, most enterprises don't get the returns. That gap is the opening NetApp is targeting with intelligent data infrastructure, framed as unified storage spanning wherever the infrastructure sits.

The Novus architecture is the concrete piece. Its Data Director manages metadata separately from stored data, and Gunna Marripudi, vice president of product management for Novus, called metadata concurrency the pillar for an architecture. The pitch extends beyond throughput: NetApp's AI Data Engine discovers, classifies and vectorizes enterprise data, aiming to cut preparation work that typically takes six to nine months of engineering.

The open-standards angle is where this becomes a practical decision for storage teams. Nair's promise is that customers don't need to rip and replace what they already have, and the planned PEAK:AIO acquisition is expected to bring pNFS-based parallel file system technology — with any Linux kernel released after 2018 already carrying a pNFS client. Whether that lands as advertised appears to hinge on how smoothly these pieces slot into existing ONTAP environments.