Capabilities

NeuralMesh™ by WEKA® organizes its capabilities into five key components—Core, Accelerate, Deploy,
Observe, and Enterprise Services—that work together to fuel high-performance AI at any scale.

Core

Resilient architecture that improves with scale

Software-Defined Architecture

Deploy anywhere on a software-defined architecture that delivers high-performance storage.

Multitenancy & Isolation

Secure operations with ironclad tenant isolation and role-based access controls that are efficient and compliant.

Integrated Data Protection

Protect your data with patented erasure coding and rapid file-level rebuilds—without slowing performance.

Cloud Auto Scaling

Scale resources dynamically—up or down—to match your needs and optimize costs.

Flexible Failure Domains

Recover quickly by localizing rebuilds for fast, resilient recovery and zero performance loss, even during multiple failures.

Load Balancing

Distribute workloads intelligently and maintain consistent, resilient, and high-speed performance at any scale.

Modern Data Structures

Eliminate metadata bottlenecks with parallelized, hash-based structures for massive directory scale and concurrency.

Accelerate

Delivers unmatched I/O performance and GPU efficiency

Zero-Copy Architecture

Read and write data directly from storage—no duplication, no silos—boosting throughput while reducing latency.

Zero-Tuning Performance

Optimize workload performance automatically with no manual tuning, for any access pattern or file size.

Virtual Metadata Servers

Distribute metadata with dynamic rebalancing across all nodes to ensure low latency—even with billions of files.

Flexible Protocol Support

Unify access with built-in POSIX, NFS, SMB, and S3 support in a single namespace ensuring consistent high performance.

POSIX Kernel Bypass + SPDK/DPDK

Bypass the kernel to remove interrupt latency and ensure ultra-low I/O latency.

Load Balancing Writes

Stripe writes across failure domains and route them to the fastest nodes—enabling parallel, fast, and single-hop I/O.

Adaptive Caching

Accelerate small file performance with local page and metadata caching while keeping data consistent everywhere.

Deploy

Cloud-native, containerized, and composable by design

Fully Containerized Microservices

Run every function as an independent container for modular upgrades, rapid scaling, and granular resource assignment.

Run Anywhere Architecture

Deploy anywhere—bare metal, cloud, or hybrid—with identical software and performance.

Cloud Portability

Achieve consistent performance and features across AWS, Azure, GCP, OCI, Neoclouds, or on-prem—no refactoring needed.

Observe

Deep telemetry and real-time visibility across environments

Telemetry & Management

Monitor cluster health, resource use, and performance in real time with built-in telemetry and observability tools.

Logs, Traces, & Metrics

Track structured logs, distributed traces, and metrics for comprehensive monitoring and fast troubleshooting.

Snapshot Difference & Change Tracking

Identify changes instantly across snapshots—no file tree walking needed—for audit, DR, and data cataloging.

Latency-Aware Rebalancing

Automatically optimizes performance using latency histograms to rebalance load and move metadata processes across nodes.

Enterprise Services

Integrated capabilities for resilience, compliance, and efficiency

Snapshots and Clones

Create instant, space-efficient snapshots and clones to support restore, migration, and dev/test pipelines without degrading performance.

Snap-to-Object & Remote Backup

Replicate snapshots incrementally to object storage for fast, cost-effective backup, DR, and hybrid cloud restore.

Integrated Tiering

Tiers cold data to object storage while keeping hot data and metadata on NVMe to ensure top performance.

Encryption & Key Management

Encrypt data end-to-end with FIPS 140-3 Level 1 and integrated key management—at rest, in motion, and in the cloud.

Data Reduction

Reduce your storage footprint up to 10x with inline compression and deduplication—without impacting performance.

Be First To Test Drive NeuralMesh

Join the waitlist for a preview.