What It Looks Like at The End Of The HPC Storage Stack

SciNet, one of Canada's primary academic supercomputing centers, recently deployed a new infrastructure platform built around AMD CPUs, H100 GPUs, NDR InfiniBand, and VAST Data. The deployment provides a useful look at how modern HPC environments are evolving as traditional simulation workloads increasingly share infrastructure with AI. In discussing the design, operation, and future direction of the system, CTO Daniel Gruner highlights the growing importance of data platforms that can simultaneously deliver performance, operational simplicity, security, and multi-tenant services across a diverse scientific computing environment.