High-performance AI storage

Flash-optimized object storage feeds GPUs

Scality RING XP eliminates storage bottlenecks with flash-optimized, low-latency performance that
keeps GPUs running at full utilization through every stage of the AI pipeline.

REQUEST DEMO

Flexibility is the ultimate flex

Why next-gen AI and cloud data demand multidimensional scaling

DOWNLOAD NOW

AI data challenges driving high-performance storage needs

Keeping GPUs fully utilized

Conventional storage can’t deliver the IOPS needed to feed GPUs at full speed

Removing latency bottlenecks

Training and inference workflows demand microsecond access to data and embeddings

Bridging file and object workloads

AI teams need seamless integration between fast file systems and scalable object stores

Where performance matters in the AI data pipeline

AI pipelines depend on consistent, ultra-fast performance — especially in the stages where GPUs are most active. Scality RING XP accelerates these critical phases to keep training and inference running at full speed.

Preparation

Hot working data staged on NVMe delivers the throughput GPUs require for rapid model setup and loading

Training

Extreme flash performance keeps GPUs fully utilized during model training

Checkpointing

Frequent, performance-intensive passes over specialized datasets stay fast and reliable

Inference

Model weights, embeddings, and vector databases are served at microsecond latency for real-time, context-rich responses

Meet Scality RING XP: AI storage for extreme performance at scale

To meet the demands of AI model training and fine tuning, Scality RING XP is an AI-optimized, all-flash configuration of industry-leading Scality RING. It takes performance to the unheard of level of microsecond response-time latencies for small (4KB) object data, making it the world’s fastest object store.

RING XP is the ideal ultra-fast object storage target for AI tools, custom-developed applications, and performance-optimized file systems used for training AI models.

This provides application developers with a streamlined object storage solution that provides:

Fast object storage API-based access
Unbounded scale to billions of objects and beyond
Extreme performance for small object data with microsecond access latencies
Easy-to-manage and affordable solution with intuitive UI dashboard
Choice of AMD EPYC-based storage servers

WHY SCALITY RING XP FOR HIGH-PERFORMANCE AI APPLICATIONS

Extreme performance and scalability at a fraction of the cost

Grow without limits

Encounter zero limits in number of objects, storage capacity or performance. Tested and proven at massive scale with the world’s largest customers.

Ultra-low latency

Extreme performance for small object data with microsecond access latencies, making it the perfect solution for AI inferencing, real-time analytics, and other high-demand, transaction-heavy environments.

Choice of durability + availability

RING XP provides all the durability advantages of RING (erasure coding, replication, self-healing), or optional reduced durability policies for extreme storage performance.

API-based access

RING XP provides a simple, ultra-fast and streamlined object storage API for application developers. This fits naturally with the patterns of cloud-native and AI applications.

Support for all data types

AI data in a wide variety of forms are supported on RING XP, whether they are IoT event streams, logs, documents, images or other.

Scale-out throughput

For large object data, including images and videos, RING XP can be scaled out to provide ultra-high levels of throughputs beyond GB/sec.

Control + sovereignty

Deploy RING XP in the geography and data center of your choice to ensure sovereignty, enabling close proximity of apps to storage for optimal performance.

Low cost

Licensed via affordable, usable capacity-based subscription model, RING XP leverages commercial storage servers based on AMD EPYC processors and NVMe flash drives. Efficient use of hardware resources further drives down costs.

Integrated with the AI ecosystem

Scality RING XP works with the platforms enterprises already use

Enterprises don’t build AI in isolation. Scality integrates with the frameworks, accelerators, and orchestration tools already powering today’s AI workloads, ensuring compatibility from day one.

WEKA + RING XP

Joint flash solution delivering up to 10× faster throughput than traditional S3 — proven for GPU-optimized training pipelines.

Certified ISV integrations

More than 20 tools validated across the AI lifecycle

The Scality advantage

Scality RING was designed to solve the same scale, performance, and resilience pressures that AI now pushes to the extreme

MultiScale

Seamlessly grows across capacity, performance, and operations — from terabytes to exabytes

CORE5

Five layers of cyber-resilience defend datasets against ransomware, corruption, and failures

Adaptability

Freedom to choose hardware, media (HDD, SSD, NVMe), and deployment models without lock-in

Extreme performance for AI, proven in production

From GPU-driven training to real-time inference, Scality RING XP delivers the performance modern AI pipelines demand. By combining flash-optimized speed with CORE5 resilience, RING XP ensures GPUs stay fully utilized and AI results arrive without delay.

BOOK A DEMO NOW