Slurm, native to Kubernetes
Slurm's scheduling, running as Kubernetes workloads. Users keep the sbatch, srun and sacct they know; operators get Kubernetes' elasticity and automation.
Storm
Ready when your users are.
Run HPC and AI workloads on any cloud or in your own data centre: the Slurm scheduler your users already know, a desktop in the browser, CPU and GPU workloads side by side, and the budget control and security your organisation needs. Built by nAG.
Free 4-hour trial: a Slurm desktop in your browser. Sign up with an email address — nothing to install.
At the core, Slurm runs natively on Kubernetes through Slinky, the open-source project by SchedMD. Around it, nAG adds what an organisation needs to run it as a platform.
Slurm's scheduling, running as Kubernetes workloads. Users keep the sbatch, srun and sacct they know; operators get Kubernetes' elasticity and automation.
Google Cloud, AWS, Azure or on-premises. Elastic CPU and NVIDIA GPU node pools that start when jobs need them and scale to zero, spot tiers for throughput.
Shared POSIX storage on every node, parallel filesystem and object storage integrations, hybrid and cross-cloud data movement.
Deployed into the customer's own cloud organisation and network, under their policies: single sign-on, customer-managed keys, egress control — everything as code.
Curated Spack software stacks, Intel oneAPI and NVIDIA CUDA toolchains, the nAG Library, and nAG's HPC specialists for onboarding and performance.
A full Linux desktop in the browser, GPU-accelerated visualization, a web file manager, and a guided tutorial to the first Slurm job.
Per-user and per-project budgets enforced by Slurm, showback per job, and scale to zero and idle shutdown by default.
Cluster, queue and job metrics, usage analytics and alerting, and an audit trail of who ran what, where and when.
nAG builds Storm in your cloud or data centre, and your team runs it.
nAG operates Storm for you, in your environment.
Storm as a service: sign in and run.
Try the live demo, walk through the architecture, and tell nAG what your HPC platform needs.