Backend.AI vs Slurm Workload Manager

Backend.AI is an AI infrastructure platform for building, training, and serving models, as well as running parallel computing jobs. Its container-level GPU virtualization divides physical GPU capacity into fractional allocations so multiple users can share a GPU. The Sokovan scheduler supports multi-node, multi-tenant workloads and policy-based resource management. The platform says its hardware abstraction layer supports more than 12 accelerator types, including products from NVIDIA, AMD, Intel, Rebellions, and FuriosaAI. It can be deployed on-premises, in public clouds, or in a hybrid setup; installation, model deployment, and updates can operate without internet connectivity in air-gapped networks. The browser-based WebUI provides cluster management, resource and session monitoring, and policy-based allocation. Listed cloud environments include AWS, GCP, Azure, and OCI. Storage integrations include VAST, WEKA, Pure Storage, and IBM Storage Scale, with NVIDIA GPUDirect Storage support. Open-source is free for home users installing on their own hardware. Enterprise plans require a quote. Production guidance says compute nodes should be network-isolated with restricted inbound access.

Slurm Workload Manager is free, open-source software for allocating compute resources and scheduling work across Linux clusters. It starts and monitors jobs on allocated nodes and manages queues of pending work. Optional plugins add accounting, advanced reservations, gang and backfill scheduling, topology-aware resource selection, resource limits, and multifactor job prioritization. The system uses a central slurmctld manager and slurmd daemons on compute nodes; accounting and REST components are optional. GPU plugins support NVIDIA, AMD, Intel, AWS Trainium, and AWS Inferentia accelerators, along with NVIDIA MPS and GPU sharing. Accounting can use MySQL or MariaDB and can record data in a text file, directly in a database, or through slurmdbd for multiple clusters. Slurmrestd provides REST access, but the API is not designed for direct internet exposure and supports only unencrypted HTTP. Linux is thoroughly tested on arm64, ppc64, and x86_64; FreeBSD and NetBSD support is limited, and macOS is unsupported. Deployment is self-hosted.