Slurm Workload Manager
slurm.schedmd.com
- Free plan
- Linux
- API
- Self-hosted
What Slurm Workload Manager is
Slurm Workload Manager is free, open-source, self-hosted software for allocating resources and scheduling jobs across Linux clusters. It starts and monitors work on assigned compute nodes and manages queues of pending jobs. Optional plugins add accounting, advanced reservations, gang scheduling, backfill scheduling, topology-aware resource selection, resource limits, and multifactor job prioritization. The architecture centers on slurmctld, with slurmd daemons on compute nodes and optional slurmdbd accounting and slurmrestd REST access. GPU plugins support NVIDIA, AMD, Intel, AWS Trainium, and AWS Inferentia accelerators, plus NVIDIA MPS and GPU sharing. Accounting can use MySQL or MariaDB, text files, databases, or slurmdbd for multiple clusters. SchedMD documents integrations for Google Cloud, AWS Parallel Computing Service, Microsoft Azure, CycleCloud, and Kubernetes interoperability through Slinky. Linux is thoroughly tested on arm64, ppc64, and x86_64; FreeBSD and NetBSD have limited support, while macOS is unsupported.
Who it suits
Slurm suits teams running high-performance computing, AI workloads, and large or small Linux clusters. It is intended for organizations able to deploy and administer self-hosted cluster software.
Good to know, and what to watch
- Free GNU GPL v2 software
- Allocates resources and manages job queues
- Optional scheduling and accounting plugins
- GPU support across several accelerator families
- Cloud and Kubernetes interoperability options
- Self-hosted deployment required
- macOS is not supported
- REST API supports only unencrypted HTTP
- REST API is not designed for direct internet exposure
Plans
Every plan and priceSlurm
Free plan
The verdict
Slurm provides broad cluster scheduling, accounting, GPU, and integration options without a monetary license cost. Its self-hosted Linux focus and REST security limitations matter when planning deployment.
What the maker publishes
- Architecture
- Slurm uses a centralized slurmctld manager, slurmd daemons on compute nodes, optional slurmdbd accounting, and optional slurmrestd REST access. source
- Authentication
- Slurm authenticates communications between components through a configurable authentication plugin, with MUNGE as the default and recommended option and an internal auth/slurm plugin also available. source
- Cloud and Kubernetes
- SchedMD documents Slurm integrations for Google Cloud, AWS Parallel Computing Service, Microsoft Azure and CycleCloud, while Slinky provides interoperability between Slurm and Kubernetes. source
- Core functions
- It allocates compute resources, starts and monitors work on allocated nodes, and manages queues of pending work. source
- Database integration
- Slurm supports MySQL or MariaDB for accounting and can record accounting data in a text file, directly in a database, or through slurmdbd for multiple clusters. source
- Founded
- 2010 source
- GPU support
- Optional GRES plugins support NVIDIA, AMD, Intel, AWS Trainium, and AWS Inferentia accelerators, plus NVIDIA MPS and GPU sharing. source
- Headquarters
- Utah, United States source
- Notable adoption
- SchedMD states that Slurm is the scheduler of choice for over half of the top one-hundred systems in the TOP500. source
- Primary audience
- SchedMD describes Slurm as a workload manager for high-performance computing and AI environments, including large-scale supercomputing systems. source
- Project history
- Slurm was initially released in 2002, and SchedMD was founded in 2010 to continue developing it. source
- Purpose
- Slurm is an open-source, fault-tolerant, highly scalable cluster management and job scheduling system for large and small Linux clusters. source
- REST API
- Slurmrestd provides a REST API, but the API is not designed to be directly internet-facing and supports only unencrypted HTTP communications. source
- Scheduling features
- Optional plugins provide accounting, advanced reservations, gang scheduling, backfill scheduling, topology-aware resource selection, resource limits, and multifactor job prioritization. source
- Supported operating systems
- Linux is thoroughly tested on arm64, ppc64, and x86_64; FreeBSD and NetBSD have limited support, while macOS is not currently supported. source
- Support services
- SchedMD offers standard support, deployment services, on-site training, and technical account manager services for Slurm and Slinky. source
Checked against the maker's pages on 1 Oct 2026.