AMPBELL, California, June 29, 2021. – WekaIO™ (Weka), one of the fastest-growing data platforms for artificial intelligence/machine learning (AI/ML), life sciences research and High-Performance Computing (HPC), today announced that its WekaFS™ unlimited data platform can leverage the new enhanced NVIDIA HGX™ AI supercomputing platform to deliver the increased performance and higher throughput demanded by the most demanding enterprise workloads.
The NVIDIA HGX platform supports all three core technologies: the NVIDIA A100 80GB GPU, NVIDIA NDR 400G InfiniBand networking, and NVIDIA Magnum IO GPUDirect® storage software.
Weka’s support of NVIDIA’s next-generation technologies is the result of close collaboration between the two companies and Weka’s ongoing commitment to providing customers with the most advanced features for the best storage, protection and data management solution.
WekaFS™ also recently outperformed all other storage systems tested by Microsoft Research, with the best overall performance of all NVIDIA Magnum IO GPUDirect® storage systems. In testing a system in which WekaFS was deployed with multiple NVIDIA DGX-2™ servers in a service delivery environment, engineers found that WekaFS provided the best performance of any storage solution tested to date. This high performance was achieved and verified by running the NVIDIA HGX™ tool as a stress test and showed consistent performance throughout the test.
Different phases of AI data channels have different storage requirements due to the huge input bandwidth, which requires mixed read and write management, and extremely low latency, often resulting in storage silos for each phase. This means business and IT leaders need to rethink the way they design storage systems and make purchasing decisions for new workloads. By taking advantage of NVIDIA platform enhancements, customers can better take advantage of significant performance gains with significantly less overhead, process data faster while minimizing power consumption, and scale AI/ML workloads by improving accuracy.
The use of WekaFS, along with new additions to the NVIDIA HGX AI supercomputing platform, will enable partners to deploy high-performance solutions in a growing number of industries that require fast access to data – both on-premises and in the cloud. The industries they serve include manufacturing, life sciences, scientific research, energy exploration and extraction, and financial services, where HPC storage is needed to meet application demands.
WekaFS is the world’s fastest and most scalable POSIX-compliant parallel file system designed to overcome the limitations of traditional file systems using local, NFS or block storage, making it ideal for AI-intensive workloads and high-performance computing. WekaFS is a clean design that integrates NVMe-based high-performance flash storage with GPU servers, object stores, and interconnects with extremely low latency in an NVMe-over-Fabrics architecture, creating a scalable storage system with extremely high performance. WekaFS performance scales linearly as servers are added to the storage cluster, allowing the infrastructure to adapt to growing business needs.