Towards Real-Time Classification of HPC Workloads via Out-of-band Telemetry

This paper describes the first work done towards my PhD. In it, I show that out of band data avilable via Redfish may be sufficient to perform classification on HPC workloads – in real time. This has important applications in security – specifically around detecting malicious workloads and misuse of HPC systems.

I presented the paper at the HPCMASPA Workshop at IEEE Cluster 2022. The slides I used are avilable as odp, pptx, or pdf.

Code is available on GitHub.

(Per IEEE Policy, this is a preprint version. As of the time of writing, the full version was not yet on the IEEE Xplore website)

Download pdf