NetApp rolled out version 12 of its StorageGRID object based storage platform this week, saying the update will mean improved scaling for AI workloads, while advanced caching will boost performance on training and HPC workloads by 20 times.
Customers will be able to version their AI datasets with bucket branches, with support for “space-efficient clones of their object storage buckets to use in their workflows.”
Those space efficient clones will also enable faster recovery from disruptions, NetApp said. Other resiliency features include “enhanced encryption standards with AES GCM encryption add integrity checking, stronger on-disk encryption, and default blocking for SSH ports.” Support for object lock in cross-grid replication makes for easier immutable datasets for AI.
Unsurprisingly, NetApp is a strong advocate for object storage as the answer to maintaining an AI content repository at scale.
“These datasets are extremely large, and customers…often have teams of developers working on these datasets in parallel,” wrote director of product management Vishnu Vardhan in a blog accompanying the announcement. “This part of the AI workflow has been in dire need of a simple and scalable solution.”
But traditional tools either didn’t scale, changed data formats, or changed the way applications interacted with storage, he continued.
StorageGRID would now allow devs to “make instant copies of large buckets with billions of objects and petabytes of capacity, operate on these buckets independently of each other, and reconcile changes between buckets.”
An integrated cache would simplify the use of caching in AI workflows, Vardhan wrote, tighten up security, and provide consistency. This would deliver up to “10 times the performance of current NetApp StorageGRID appliances.” And this could be scaled up further by running the caching layer on a bare-metal StorageGRID node.
Vardhan added that it had doubled the “published limits” for StorageGRID and the platform can now support over 600 billion objects in a single full-size StorageGrid cluster.
In its most recent results, NetApp CEO George Kurian said it had scored 125 AI wins in its most recent quarter, compared to 50 a year earlier, and hinted it was making progress in the hyperscaler AI market.
NetApp’s announcement was fired out the same day a trio of NetApp veterans popped up at VAST Data, including former CTO and SVP Jonsi Stefansson, just months after they left their erstwhile employer.
Stefansson, now general manager for Cloud Solutions at VAST, said in a statement: “Having competed against VAST and lost more times than I’d like to admit, I knew from the outside just how strong and unique VAST’s architecture is – it’s so much more than a file system or a simple storage operating system.”
