DDN recently announced it has secured the number one position in the IO500 benchmark, a key industry metric assessing storage performance under typical Artificial Intelligence (AI) and High-Performance Computing (HPC) workloads. The IO500 ranking measures data storage technologies under realistic operational conditions, including mixed workloads, complex I/O patterns, intensive metadata tasks, and concurrent operations.
The IO500 benchmark places DDN at the top specifically in the 10-node production category, commonly representative of real-world AI workloads. DDN reports it delivers between three and 11 times greater performance for AI training, simulation, and analytics compared to competing vendors such as Weka and VAST. DDN’s architecture provides multi-terabyte throughput, high concurrent metadata processing capability, native parallel file access optimized for AI applications, pipeline-aware caching to reduce latency, and resilience suitable for enterprise deployments.
DDN identifies target data center markets for this technology deployment as finance, genomics research, and supercomputing. For example, a global hedge fund accelerated algorithm development by three times while achieving significant infrastructure cost reductions; genomics researchers at TGen reduced analytics pipelines from 12 hours to under two; and supercomputing organizations such as CINECA, Helmholtz Munich, and Bitdeer AI have reported improved GPU utilization and lower total cost of ownership (TCO).
“Being ranked #1 on the IO500 benchmark is more than a technical achievement — it’s proof that our customers can count on DDN to deliver the speed, scale, and reliability needed to turn data into competitive advantage,” said Alex Bouzari, CEO and Co-Founder of DDN. “DDN is not just ahead. We have left the competition behind.”
Competitor analysis published by DDN indicates that Weka storage solutions require manual tuning and have limitations under concurrent workload conditions. VAST storage systems reportedly struggle with write-intensive and metadata-heavy workloads, while Hammerspace lacks relevant deployments and IO500 benchmark trials.
Full benchmark results and additional details are available on the IO500 website: io500.org.
Source: DDN
