JC John Cuckovich
All work Get in touch
CASE STUDY / 03

NVIDIA vGPU on VCF

GPU acceleration on VMware Cloud Foundation — automated from the BIOS up, with the vGPU lifecycle owned end to end.

ROLE  Lead Network Engineer
WHERE  Lumen Technologies
WHEN  2018 — 2026
STACK  VCF 5.2+ · L40S · Redfish · ESXi · Aria
03

GPU acceleration, automated from the BIOS up.

VCF 5.2+ NVIDIA vGPU Redfish L40S

The situation

Adding GPU acceleration to a VCF environment isn't a single toggle — it reaches from firmware settings on the physical host all the way up through the hypervisor and management stack. Doing it by hand across a cluster is exactly the kind of error-prone, hard-to-repeat work that undermines a platform. The target was an L40S GPU cluster running vGPU on VCF 5.2+.

What I built

I delivered NVIDIA vGPU acceleration on VCF 5.2+ for an L40S GPU cluster deployment. I personally automated BIOS provisioning via Redfish — configuring the hardware out-of-band before the host ever booted an OS — plus the ESXi, vSphere, and VMware Aria layers above it, so the whole GPU-ready stack came up the same way every time.

Automating from the BIOS up meant the physical-to-virtual handoff stopped being a manual checklist and became part of the pipeline.

GPU on a private cloud lives or dies on the boring details — firmware, driver, and manager versions all agreeing with each other.

Owning the lifecycle

I served as SME for NVIDIA vGPU Manager lifecycle and compatibility validation — the ongoing work of keeping firmware, host software, and vGPU Manager versions aligned so the acceleration stayed supported and stable as the platform moved forward.

← PREVIOUS
HashiCorp Vault, Highly Available
NEXT →
CI/CD Modernization
← Back to all work jpcuckovich@gmail.com ↗