Scality and WEKA introduce NeuralMesh connector for RING

Scality and WEKA announced a jointly validated solution that combined WEKA’s NeuralMesh flash storage with Scality RING’s object tier to address performance and capacity needs for Artificial Intelligence (AI) and High performance computing (HPC) workloads.

Scality reported test results showing its RING interface delivered up to 10x the speed of conventional S3 interfaces on similar hardware and reduced infrastructure costs by up to 20 percent; the companies said those outcomes enabled more economical extension of NeuralMesh-powered AI and HPC data pipelines while avoiding the management overhead associated with traditional object stores.

At the center of the integration was Scality’s object connector for NeuralMesh, described in the release as a lightweight object storage interface that WEKA validated at scale; the joint architecture kept new and active data on NeuralMesh’s flash tier and tiered other data to Scality RING, which the companies said scaled to exabytes with up to 14 nines durability.

The release listed customer benefits that included higher Graphics Processing Unit (GPU) utilization, faster time to first token, flash-speed AI pipelines, an object tier the companies said could lower costs by up to 20 percent compared with traditional object integration, validated interoperability that required no engineering changes, and a connector the companies described as faster and more manageable than community-driven object stores.

“WEKA’s NeuralMesh storage system delivers the high-performance software foundation modern AI pipelines require to run optimally. Enterprises that leverage the Scality RING lightweight connector together with NeuralMesh can achieve additional economic benefits, leveraging a cost-efficient object tier,” said Nilesh Patel. “Our partnership with WEKA gives enterprises a smarter way to extend the economics of their AI pipelines,” said Erwan Girard. The companies said the Scality RING with NeuralMesh Connector for WEKA solution is available now.