Industry : Multinational; Medical Diagnostic Equipment

Objective :  Automate the provisioning of vGPU enabled VMs for test/dev activities in ML workloads, reducing setup time from half a day to an automated process.

Challenges : Required a custom vRealize Automation blueprint and orchestrator workflows, with difficulties including special hardware, vGPU knowledge, various hardware models, and limited API support.

Solutions

Developed a custom workflow for VM placement and load balancing across clusters, with automatic vGPU profile allocation.

Benefits

Reduced deployment time from 4 hours to 1 hour, allowed business user-driven requests, and simplified workload tracking.