Scality Ecosystem & Application Partners
Scality and WEKA NeuralMesh.
Flash-speed for active AI data, scale-out object for everything else. WEKA delivers the performance; Scality extends the economics. The joint solution is the Scality RING Connector for NeuralMesh by WEKA.
the problem
AI and HPC workloads keep outgrowing what all-flash can afford.
Enterprises building AI and HPC pipelines hit the same wall. Models and datasets grow faster than budgets, and most architectures force a choice between performance and cost.
- Model and dataset sizes grow faster than infrastructure budgets.
- All-flash storage does not scale economically beyond the active set.
- Active datasets still need flash-speed access. Cold datasets still need to be reachable.
- Object stores bolted onto file pipelines add latency and operational overhead.
the joint solution
The Scality RING Connector for NeuralMesh removes that choice.
WEKA holds the performance tier. Scality RING holds the capacity tier. A lightweight REST-based connector — validated by WEKA — joins the two.
- WEKA NeuralMesh: flash-speed file for active AI and HPC data.
- Scality RING: durable, exabyte-scale object storage for everything cold.
- Active data stays on WEKA. Cold data tiers seamlessly to Scality.
- No all-flash forced by the storage layer. No engineering changes on the WEKA side.
joint solution benefits
What flash and object together deliver.
Performance where it matters.
NeuralMesh maximizes GPU utilization, accelerates time to first token, and powers AI pipelines at flash speed. Active data never waits.
Economics at scale.
The RING Connector cuts infrastructure cost 5–20% versus traditional object integration. Cold data lands on an object tier sized for petabyte-to-exabyte budgets.
Proven interoperability.
Validated by WEKA. The connector is REST-based and ships as a standard option in the WEKA configuration. No engineering changes required.
A reliable alternative to Ceph.
Enterprise-grade support on both ends. None of the operational tax of community-driven object stores.
Joint architecture
How WEKA NeuralMesh runs on Scality.
Workloads flow through the partner integration and land on the Scality data layer.
Show us the workflow. We will show you the fit.
