Storage for zettascale systems is an area of significant challenge. The explosion in the production and use of data in science, engineering, medicine and industry has led to performance bottlenecks and system slowdowns as traditional high performance computing storage architectures fail to keep up with the increasingly important role that analysis plays in supercomputing workflows.
A new approach to storage for exascale is required, incorporating state-of-the-art devices such as NVMe disks and Intel® Optane™ Data Center (DC) Persistent Memory. These promise to revolutionise I/O performance when integrated into new storage subsystem designs.
In collaboration with Dell Technologies, Intel and the University of Cambridge, the Zettascale Lab has a wealth of experience designing, implementing and operating large-scale storage solutions for high performance computing, with a focus on solid-state technologies.
One example is the Cambridge Data Accelerator, which is currently the UK’s fastest high performance computing storage solution. It also claimed first place in the IO-500 worldwide high-performance computing storage performance ranking of June 2019.
Aims
The First ExCALIBUR International Extreme Data (EIED) Workshop will take place from July 18 – 21 in Cambridge, UK.
Principal Systems Engineer at COZL Bob Cregan speaks on the topic of integrating a high performance burst buffer with the slurm resource scheduler.
Our ExCALIBUR International Extreme Data programme kicked off on July 18 2023.