NetApp introduces Novus AI storage architecture

NetApp has unveiled Novus, a storage architecture designed to support large-scale AI factory operations. The solution targets AI infrastructure providers, such as neocloud companies and GPU-as-a-Service providers, by aiming for an aggregate throughput of over 100 TB per second to prevent GPU underutilization.

To address data bottlenecks, Novus utilizes an approach that separates metadata management from the data path. This design allows performance, capacity, and concurrency to be scaled independently within a single namespace via the NFS protocol. In its initial phase, Novus integrates the NetApp Novus Data Director on qualified Supermicro infrastructure with NetApp ONTAP data services via the AFF A90 system.

Additionally, NetApp announced new hybrid multi-cloud capabilities featuring autonomous controls. These updates aim to provide consistent data strategies across on-premise, public cloud, and sovereign regions. Features include NetApp Keystone Sovereign, which supports data sovereignty compliance in the European Economic Area, and NetApp Console Autonomous Operations, which uses automation to reduce reactive workloads.