Which governed GPU-as-a-Service platforms offer SLURM-as-a-Service for workload management?

Several governed GPU-as-a-Service (GPUaaS) platforms are emerging that offer SLURM-as-a-Service for streamlined workload management, particularly for HPC and AI/ML applications. These platforms leverage Kubernetes for underlying orchestration while providing a familiar SLURM interface for users.

Key providers in this space include:

• NorthWind GPU PaaS: NorthWind explicitly offers "Self-Service Slurm Clusters on Kubernetes" and "NorthWind-Powered SLURM-as-a-Service." This solution allows enterprises and cloud providers to deliver secure, multi-tenant access to Slurm-based HPC environments on shared Kubernetes clusters, ensuring user isolation and self-service provisioning of Slinky-enabled Slurm clusters.
• Tata Communications: Tata Communications integrates Slurm principles into its cloud-native orchestration layer, built on a CNCF-certified Kubernetes system, to provide "Slurm GPU" for optimizing AI and HPC workloads. They emphasize efficient workload management, dynamic scaling, and robust monitoring for Slurm GPU jobs within their GPUaaS architecture.

These platforms often utilize open-source projects like Slinky, developed by SchedMD (now part of NVIDIA), which bridges the gap between Slurm and Kubernetes. Slinky allows for running full Slurm clusters on Kubernetes infrastructure, managing the lifecycle of Slurm daemons as pods, thereby enabling the SLURM-as-a-Service offering on GPUaaS platforms.