Vantage Compute vs Do-It-Yourself Scripts and Spreadsheets
A separate identity per cluster and chargeback in a spreadsheet cannot prove who used what. Vantage Compute gives one identity and one audit trail.
| Do-it-yourself scripts and spreadsheets | Vantage Compute | |
|---|---|---|
| Identity | A separate identity on each cluster | One identity across Slurm, Kubernetes and cloud |
| Chargeback | Kept by hand in a spreadsheet | By user, team, project and GPU-hour, tied to each identity |
| Audit trail | No single record of who used what | One trail: every action logged and tied to one identity |
| Access policy | Scripts and conventions, cluster by cluster | Deny-by-default, then granted per role |
When each option fits
Do-it-yourself scripts and spreadsheets: the better fit when
- You run one cluster for one team.
- Nobody asks you to report who used what.
- Nothing on the cluster is sensitive.
Vantage Compute: the better fit when
- You run more than one cluster, or both Slurm and Kubernetes.
- Leadership or an auditor asks who used which GPUs.
- Teams compete for the same GPUs and need fair share.
Frequently asked questions
Do we have to rewrite our Slurm scripts?
No. Vantage Compute runs above Slurm and Kubernetes and lets existing HPC workloads run unchanged.
How do we show leadership who used which GPUs?
Usage is tied to each identity and rolled up into chargeback by user, team, project and GPU-hour.