Scarus
An OCI-compatible container engine designed specifically for running Linux containers on High Performance Computing (HPC) environments.
What is Scarus?
Sarus is an OCI-compatible container engine specifically designed for High Performance Computing environments. It enables researchers and engineers to run Linux containers on HPC systems while addressing the unique security, performance, and compatibility requirements of supercomputers and clusters. The software allows users to deploy containerized applications with native performance on custom hardware through extensible runtime hooks.
HPC system administrators, researchers, and engineers who need to run containerized applications on supercomputers, clusters, and other high-performance computing environments with specific security and performance requirements.
Developers choose Sarus because it's specifically engineered for HPC environments, offering security hardening for multi-tenant systems, compatibility with workload managers, and optimized performance for parallel filesystems—features not typically found in general-purpose container engines.
Overview
OCI-compatible engine to deploy Linux containers on HPC environments.
Use Cases
Best For
- Running containerized scientific applications on HPC clusters
- Deploying applications on diskless HPC nodes with parallel filesystems
- Securing container execution in multi-tenant supercomputing environments
- Integrating containers with existing HPC workload managers
- Achieving native performance with custom hardware through OCI hooks
- Building reproducible research environments on HPC infrastructure
Not Ideal For
- Deploying containerized web applications or microservices on standard cloud infrastructure without HPC requirements
- Small teams or individual developers working on local machines or small servers without parallel filesystems
- Projects that rely heavily on Docker Compose, Kubernetes, or other orchestration tools not integrated with HPC workload managers
Pros & Cons
Pros
Implements a security model specifically tailored for multi-tenant HPC systems, addressing unique risks in supercomputing environments as highlighted in the key features.
Uses extensible OCI runtime hooks to support custom hardware like GPUs and accelerators, enabling containers to achieve near-native performance on HPC infrastructure.
Creates container filesystems optimized for diskless nodes and parallel filesystems such as Lustre or GPFS, crucial for high-performance I/O in clusters.
Designed for compatibility with HPC workload managers like Slurm or PBS, ensuring seamless job scheduling and resource management alongside containers.
Cons
Its specialization limits applicability outside HPC environments, with fewer features for general-purpose container orchestration and a steeper learning curve for non-HPC users.
Requires specific HPC infrastructure and expertise for installation and configuration, making it more involved than mainstream options like Docker or Podman.
As a specialized tool, it has a smaller user base and fewer pre-built images or third-party integrations, which can increase development overhead and troubleshooting time.
Frequently Asked Questions
Related Projects
HPC Container MakerHPC Container Maker
CharliecloudCharliecloud provides user-defined software stacks (UDSS) for high-performance computing (HPC) centers (Source Code) Apache-2.0
uDockerA basic user tool to execute simple docker containers in batch or interactive systems without root privileges (Source Code) Apache-2.0
ShifterShifter is Linux containers for HPC (Source Code) other
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.