Scality Ring XP All-Flash Fast Object Store for High-Performance AI Applications

Scality SA unveiled RING XP, an all-flash configuration of its RING storage software engineered for extreme performance.

RING XP is a fast object store with microsecond response-time latencies for small object data, making it for AI model training and fine-tuning.

It is 10–20x faster than Amazon S3 Express One Zone and 50–100x faster than Amazon S3 in terms of latency (milliseconds compared to microseconds). It is a 1st software-defined object storage solution to perform at levels previously only available from most costly, proprietary, hardware-based all-flash file systems and block arrays – but with the inherent advantages in scalability, simplicity, API access, security and cost provided by object storage.

RING was recently ranked as #1 on the 2024 GigaOm Radar for Enterprise Object Storage – achieving the highest scores across key features, emerging features and business criteria categories, well ahead of 17 competing vendors. It already delivers scale-out, high-throughput capabilities used by 100s of enterprises and service providers worldwide to power their critical business workloads.

“We have evolved the world’s most comprehensive and flexible object store and made it the world’s fastest,” said Giorgio Regni, CTO and co-founder. “Organizations globally have been looking to maximize how AI delivers value for their business. Object storage is a natural foundational repository for exabytes of data across the AI pipeline. With RING XP, we’ve not only optimized object storage for extreme performance but also reduced data silos – offering one seamless flexible technology for both long-term data retention and ultra-high performance AI workloads.“

RING XP: eXtreme Performance solution for AI model training and fine-tuning with AI applications

RING XP can scale to EBs and provide microsecond latency
To take object storage performance to the new level of ms response-time latencies, RING XP is based on:

• AI-optimized RING XP object storage connectors to provide scale-out, fast access to storage from applications.
• A performance-tuned RING storage software that accelerates storage I/O throughout the stack.
• AMD EPYC-based all-flash NVMe storage servers from Lenovo, Supermicro, Dell and HPE. EPYC provides support for PCIe and NVMe, and the highest number of cores in single-socket CPUs to deliver optimal latencies.

By achieving ms-level write (PUT) and read (GET) latency for 4KB objects, RING XP delivers performance to make it fast object storage target for AI tools, custom-developed applications, and performance-optimized file systems used for training AI models. Additional details on how RING XP attains microsecond latencies – considered the world’s fastest for object storage – are documented in the Solved blog.

RING XP, together with RING, delivers an end-to-end AI data pipeline storage management solution, designed to optimize and accelerate emerging AI-enabled business processes. Unlike traditional storage solutions that address only fragments of the AI workflow, they provide a unified platform that supports every stage of the AI pipeline – from ingesting massive datasets to model training, inference, and beyond.

Solution provides:

• RING as the foundational repository for AI data lakes with warm/cool/cold data at exabyte scale
• RING XP for the extreme low-latency requirements of AI training data sets
• Integrated lifecycle management for all phases of the data pipeline
• Managed through a common framework of management and monitoring tools
• End-to-end cyber-resiliency through integrated CORE5 capabilities to ensure security and data privacy

RING is the AI data lake repository of choice for corporations in financial services, genomics, utilities, intelligence services, travel and research institutions. Customers include French genomics laboratory SeqOIA Genomique, a top-5 US bank for fraud detection, one of the largest insurers in USA for claims processing optimization, and the leading European provider of AI-driven travel services.

Mary Jander, senior analyst, Futuriom, said: “As AI increasingly enters the workloads of enterprises and governments worldwide, performance has become a vital aspect of data storage. Scality’s RING XP addresses the growing demand for a full-spectrum approach to AI storage requirements, including the need to provide exabyte-scale object storage for use in accelerated computing environments.” 

Kumaran Siva, corporate VP, strategic business development, AMD (Advanced Micro Devices Inc.), said: “AMD EPYC processors provide robust PCIe and NVMe support to deliver exceptional storage performance and I/O. We are thrilled to collaborate with Scality as they enable impressive object storage performance for emerging AI applications on AMD EPYC CPU-based systems from every leading server vendor.“

Resources:
Blog: How Scality RING XP performs 20X faster than Amazon S3 Express One Zone
Use case: Extreme performance for AI workloads