Enabling Researchers to Computationally Perform Science

Overview

SciNet, the supercomputing center at the University of Toronto, provides Advanced Research Computing infrastructure for Canadian academics. Supporting fields from ocean modeling to AI, SciNet operates national supercomputing assets, including Trillium, equipped with 240,000 cores and 250 GPUs.

Solution

SciNet adopted the VAST AI Operating System to replace legacy parallel file systems and eliminate complex storage tiering. Deploying a unified 30PB all-flash platform, SciNet established a high-performance foundation capable of feeding its CPU and GPU clusters without network saturation.

Results

With the VAST Data Platform running flat-out in production, SciNet eliminated burst buffers, achieved a 2:1 data reduction ratio, and simplified administration, allowing researchers to run ocean simulations and AI workloads seamlessly.