HPE’s new Cray Storage Systems C500 offers high-performance storage solutions for supercomputing applications, particularly important in the era of AI. The parallel file system of the C500 allows for increased performance per GB per second from each SSD over the network to and from compute nodes. The ClusterStor line, which HPE acquired from Cray in 2019, scales to tens of petabytes of SSD storage, delivering impressive speeds of up to 1.6 TBps and 50 million IOPS per rack. It is utilized in major supercomputers such as Frontera, Aurora, and El Capitan, which are known as the world’s first exascale supercomputers.
The C500 shares similarities with the ClusterStor E1000 but also introduces some differences. Instead of using a 2U24 storage controller like the E1000, the C500 implements a HPE ProLiant DL325 Gen11 server for its System Management Unit. The C500 also offers more flexibility in the number of drives it can accommodate, while using fewer controllers to manage the hard drives.
According to Addison Snell, CEO of Intersect360 Research, supercomputing applications require storage solutions that can keep up with high-performance computing elements, especially with the rise of AI and new supercomputer users. Training large models on vast data sets across multiple high-speed nodes in parallel demands high-performance storage capabilities.
HPE Cray Storage Systems C500 is set to be available from HPE and its resellers starting May 27. This new system is designed to meet the increasing demands of supercomputing applications, providing high-performance storage solutions that can match the requirements of modern supercomputing tasks.
Article Source
https://www.networkworld.com/article/2098641/hpe-launches-storage-system-for-hpc-and-ai-clusters.html