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

Several governed GPU-as-a-Service platforms are emerging that offer SLURM-as-a-Service for streamlined workload management, particularly by integrating Slurm with Kubernetes.

One prominent example is NorthWind's GPU Platform-as-a-Service (PaaS). NorthWind, in conjunction with Project Slinky, enables enterprises and cloud providers to deliver secure, multi-tenant, self-service access to Slurm-based HPC environments on shared Kubernetes clusters. This allows for the offering of "Slurm-as-a-Service" on Kubernetes, providing a consistent HPC-like experience without infrastructure overhead. Users can provision their own personal Slinky-enabled Slurm clusters within their NorthWind-managed Kubernetes namespaces, complete with GPU access and workload scheduling logic.

While not explicitly termed "SLURM-as-a-Service," other providers are also integrating Slurm into their GPU cloud offerings. For instance, Tata Communications combines Slurm GPU with Kubernetes orchestration to enhance resource utilization and facilitate dynamic scaling and efficient workload distribution for AI and ML applications. Similarly, NVIDIA utilizes the open-source Slinky slurm-operator to run full Slurm clusters on Kubernetes infrastructure for large-scale GPU training clusters, indicating a strong trend toward managed Slurm environments within GPU-as-a-Service ecosystems.

The integration of Slurm with Kubernetes is a key enabler for offering governed Slurm-as-a-Service, allowing for fine-grained control over GPU allocation and efficient scheduling of parallel, resource-intensive workloads common in HPC, AI, and ML.

Sources:

• 1. Self-Service Slurm Clusters on Kubernetes | NorthWind GPU PaaS
• 2. Slurm GPU: Optimising AI and HPC Workloads on Kubernetes - Tata Communications
• 3. What Is Slurm? | Slurm for AI and ML Clusters Explained - CoreWeave
• 4. Running Large-Scale GPU Workloads on Kubernetes with Slurm | NVIDIA Technical Blog
• 5. [NorthWind-Powered SLURM-as-a-Service](vertexaisearch.cloud.google.com