Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:


DDN Benchmarks Record SPEC SFS 2014 Performance

Today DDN announced record performance results across the entire spectrum of the SPEC SFS 2014 benchmark suite. The SPEC SFS 2014 benchmarks provide a standardized approach to comparing performance for file server throughput and response time across different vendor platforms. In the latest round of testing, DDN improved on existing benchmark results by at least 25 percent.

In the past, throughput was taken to be the key metric to compare high-performance storage solutions. In this era of AI and machine learning, we have seen a fundamental shift to a heavy mix of larger operations, small file operations and significant amounts of metadata I/O,” said Sven Oehme, chief research officer at DDN. “Building a system that delivers efficiency to our customers for all GPU workloads while simultaneously adding the ability to scale indefinitely is key to supporting rapidly expanding markets such as AI and deep learning.”

The SPEC SFS 2014 benchmarks were introduced to test five aspects of a storage solution’s performance in the number of simultaneous software builds that can be performed: the number of video streams that can be captured; the number of simultaneous databases that can be sustained; electronic design automation; environment simulation; and the number of virtual desktops that can be maintained. DDN’s success across all test parameters demonstrates the flexibility and robust performance capabilities of DDN SFA platforms.

DDN’s scalable SFA platforms serve as a powerful foundation for data-at-scale challenges. The newly published SPEC results underscore DDN’s leadership in data-intensive workloads such as analytics, artificial intelligence and deep learning applications. These applications require instantaneous and consistent access to storage under heavy load in order to reduce training time and to support real-time decision making in production environments. Customers working on artificial intelligence and deep learning are collecting mixed data types like large image files along with binary telemetry and tiny text-based metadata, so data platforms must be flexible and capable of ingesting heterogenous data types without compromise.

DDN achieved the SPEC SFS 2014 results utilizing an SFA14KX configured with 72 400GB MLC SSDs. DDN’s SFA14KX platform efficiently extracts the performance of underlying flash media and delivers it to the applications without the overhead levels of other solutions. The SFA14KX is currently shipping and is being featured at upcoming fall trade shows and expositions.

Sign up for our insideHPC Newsletter

Leave a Comment

*

Resource Links: