Amazon Web Services → Microsoft Azure
Azure equivalent of AWS Parallel Computing Service (PCS)
AWS ParallelCluster (self-managed) and AWS Parallel Computing Service (managed Slurm) map to Azure CycleCloud and CycleCloud Workspace for Slurm.
About AWS Parallel Computing Service (PCS)
| Provider | Amazon Web Services |
|---|---|
| Status | Generally available |
| What it is | Fully managed Slurm. AWS runs the Slurm controller, and you define compute node groups and queues. |
| Documentation | docs.aws.amazon.com |
| Pricing | Official pricing page |
Azure mappings (1)
| Capability | Azure equivalent | Match |
|---|---|---|
| HPC cluster management (Slurm and other schedulers) | Azure CycleCloud & CycleCloud Workspace for Slurm | Direct match |
Batch & HPC(1)
HPC cluster management (Slurm and other schedulers) Direct match
AWS ParallelCluster (self-managed) and AWS Parallel Computing Service (managed Slurm) map to Azure CycleCloud and CycleCloud Workspace for Slurm.
- Managed vs self-managed. AWS PCS runs the Slurm controller for you. On Azure, CycleCloud (and the Workspace for Slurm template) deploy a CycleCloud server and Slurm cluster in your subscription that you operate.
- Schedulers. ParallelCluster and PCS are Slurm-only (ParallelCluster dropped the AWS Batch scheduler in v3.16.0). CycleCloud supports Slurm, PBS Pro, LSF, Grid Engine, and HTCondor.
- Packaging. Workspace for Slurm comes preconfigured with PMIx, Pyxis, and enroot for containerized AI/HPC jobs. PCS and ParallelCluster integrate with FSx for Lustre, EFA, and AMIs you build.