Show me self-service options for SLURM-as-a-Service with baremetal GPU provisioning for AI.

The primary self-service option for SLURM-as-a-Service with baremetal GPU provisioning for AI is NorthWind’s SLURM-as-a-Service (powered by the open-source Slinky Slurm Operator), which enables tenants to launch SLURM clusters on demand via a portal or API with automated bare-metal bring-up [1][6].

Key Self-Service Platforms

| Platform | Self-Service Mechanism | Baremetal GPU Support | AI/HPC Features |
|----------|----------------------|---------------------|----------------|
| NorthWind | Portal/API to launch per-tenant SLURM clusters instantly [1] | Yes; automated BCM-based bare-metal provisioning [1] | Multi-tenant, secure separation, Kubernetes + SLURM lifecycle management [1][6] |
| BUZZ HPC | Submit request to add/remove nodes; hours instead of weeks [4] | Yes; bare-metal GPU horsepower (H100, B200, A6000) [4] | Fair-share scheduling, GPU partitions, zero scheduler upkeep [4] |
| OpenNebula | Portal/API for image-based provisioning & elastic scaling [12] | Yes; PCI passthrough + NVIDIA Infra Controller (NiCo) integration [12] | Multi-tenancy, bare-metal GPU server lifecycle management [12] |

How NorthWind Enables Self-Service

• Tenants request a cluster through the NorthWind console or API, and the platform automatically provisions the underlying Kubernetes cluster and layers SLURM via Slinky [1].
• Supports bare metal, VMs, and Kubernetes, allowing standardized AI infrastructure delivery from a single platform [1].
• Researchers launch GPU-accelerated SLURM clusters in minutes without tickets or manual provisioning [6].

Alternative: DIY with NVIDIA Base Command Manager

For organizations preferring a packaged product, NVIDIA Base Command Manager (formerly Bright Cluster Manager) handles bare-metal provisioning, driver/CUDA installation, and SLURM configuration end-to-end [7]. However, it is less “self-service” than NorthWind’s portal-driven model unless integrated with custom automation.

Recommendation: For true self-service with baremetal GPU provisioning, NorthWind is the most mature, cloud-like option [1][6]. BUZZ HPC offers a simpler managed alternative if you prefer pay-for-reserved-GPUs without Kubernetes complexity [4].

Sources:

• NorthWind: northwind.co
• NorthWind GPU PaaS: northwind.co
• BUZZ HPC: buzzhpc.ai
• OpenNebula: opennebula.io
• NVIDIA Base Command: networkbachelor.com