VDURA Data Platform

THE VDURA DATA PLATFORM

Modern Data Storage Infrastructure for AI and HPC

The performance of a supercomputer-class parallel file system, the economics of cloud-scale architecture, and the simplicity of modern software-defined storage. Only VDURA delivers Performance, Economics, and Simplicity in one platform.

Performance. Economics.
Simplicity.

Flash-first parallel speed

True parallel file system and NVMe-first architecture keep GPUs saturated while policy automation keeps data in the right tier.

Cloud-scale economics

Shared nothing, software defined architecture enables the use of commodity hardware servers, devices, and networking. Delivering up to a 60% lower TCO with hardware flexibility and elastic expansion.

Operational simplicity

Deploy in a day, expand in minutes, and rely on self-healing automation plus non-disruptive updates for always-on availability.

The VDURA Data Platform
architecture

VDURA Data Platform V11 fuses a fully software-defined, microservices architecture with a true parallel file system and resilient object storage. The unified design delivers flash-speed performance, enterprise-grade durability, and cloud economics across active and bulk data, purpose-built to meet every stage of the AI pipeline.

Control Plane

Director Nodes

Director Nodes orchestrate every metadata transaction, coordinate Storage Nodes and DirectFlow Clients, and maintain cluster health, recovery, and reliability. Built on high-speed networking, abundant DRAM, and NVMe flash tuned for metadata logs, they are the heartbeat of the VDURA control plane.

Each Director Node runs the VeLO metadata engine – a flash-optimized, distributed key-value store engineered for ultra-low latency across billions of operations.

Data Plane

Storage Nodes

Storage Nodes dedicate their resources exclusively to user data. Flexible configurations – either all-NVMe flash for peak performance or flash paired with HDD expansion for economical capacity – deliver a single platform for active datasets and petabyte-scale archives.

Erasure Coding and VPOD layout ensure linear scalability, resilience, and efficiency across mixed-performance workloads.

Data Access

DirectFlow Client

DirectFlow Client is VDURA’s high-performance parallel file system driver. It gives compute fleets direct, POSIX-compliant access across the global namespace – keeping GPUs and CPUs saturated while bypassing legacy bottlenecks.

With metadata handled by Director Nodes and data served directly from Storage Nodes, DirectFlow sustains balanced throughput for AI training, analytics, and hybrid workloads.

Dynamic Data Acceleration

Smart Media Placement

Intent logs, metadata, small files, and large sequential workloads route automatically to SSD, NVMe, HDD, or DRAM cache – keeping latency low without sacrificing economics.

Resilience & Integrity

Targeted Reconstruction

Only affected erasure-coded objects are rebuilt, while continuous scrubbing validates data blocks so durability stays uncompromised.

AI-Aware Automation

Hands-Off Tiering

Placement engines watch file size, access pattern, and temperature to balance flash and hybrid tiers – eliminating hotspots and manual tuning.

Efficiency

On-Node Data Reduction

Compression and dedupe run inside Storage Nodes – never on clients – so GPU and CPU resources stay focused on AI. Toggle it instantly via GUI or CLI.

Proven Reliability

Always-on resilience at scale

12×9’s durability unmatched data protection
6×9’s+ availability always-on infrastructure
Resilient architecture for AI & HPC reliability improves as the network grows

Separate Control & Data Planes

Shared-nothing architecture with separate control and data planes remove contention, maximizing throughput and latency for AI & HPC pipelines at scale.

DirectFlow Clients

Enables direct, parallel access from compute nodes to VDURA Storage Nodes, removing bottlenecks found in legacy network file systems. Fully POSIX-compliant and integrated, they deliver high-speed, cache-coherent performance across AI, ML, and HPC environments.

VeLO Metadata Engine

VeLO is a distributed, flash-optimized metadata engine that keeps billions of file operations moving, accelerating staging, checkpointing, and small-file workloads.

VPODs

VDURA storage nodes host multiple Virtual Protected Object Devices (VPODs), each sharing pooled hardware resources while remaining logically independent. Adapt the number of VPODs per server to tune throughput, usable capacity, and resilience for every AI or HPC workload without touching the physical infrastructure.

Platform advantages

Superior Performance

Delivering millions of IOPS, ensuring rapid response times.
Unparalleled throughput, enabling industry leading performance.
Exceptional density to scale with business needs.

Lowest TCO

Architected for up to 60% lower TCO versus competitors.
Reliability SLAs eliminate costly premium tiers and surprise support spend.
Streamlined operations require only 0.5 FTE to manage the entire system.

Operational Simplicity

Intuitive web-based interface orchestrates end-to-end operations.
Integrated data services and a single control plane remove swivel-chair management.
Set up new capacity and expand clusters in minutes.

Raytheon case study: 60% reduction in system administration time with VDURA.

Guaranteed Reliability

Industry-leading durability delivers up to 12 nines (99.9999999999%) data protection.
Six nines+ (99.9999%) availability keeps AI and HPC workloads online without disruption.
Multi-level erasure coding shields data at both client and node tiers.
Protection grows with scale, matching hyperscale cloud resilience.

Focus on Innovation (Not Complexity)

The VDURA Difference

Hormel Institute: safeguarding critical research during breakthrough discoveries.

VDURAbility Guarantee

VDURA Care Premier underpins the VDURAbility Guarantee with ironclad availability and durability SLAs plus 10-year coverage that includes device replacement. Your data estate stays protected, predictable, and ready for every AI and HPC workload.

Explore Our Industries

See how VDURA accelerates outcomes across sectors that demand endurance, performance, and simplicity.

Academic Research

Data-driven solutions for researchers handling diverse workloads.

Artificial Intelligence

Optimized performance for cutting-edge AI applications.

Energy

Scalable tech to support sustainable energy innovation.

Federal

Secure platforms for high-performance government computing.

Financial Services

Accelerating data access for compliance and growth.

Healthcare & Life Sciences

Powering breakthroughs in healthcare, from genomic discoveries to advanced medical imaging.

Manufacturing

Enabling efficient and robust industrial operations.