BeeGFS is a leading parallel cluster file system, purpose-built to handle I/O-intensive workloads in performance-critical environments. Organizations, universities and researchers worldwide choose BeeGFS for its ease of installation, maximum scalability, robustness and flexibility. The system raises productivity by accelerating results and enabling new methods of data analysis.
Developed by the Fraunhofer High Performance Computing Center, BeeGFS was built to overcome the “hard to learn / hard to manage”, “not designed for both small and large files” and “expensive” pain points of traditional parallel file system architectures.
How It Works
BeeGFS is software-defined storage built on the POSIX file system interface — no application rewrites required. Clients reach the file system over TCP/IP or RDMA-capable networks (InfiniBand, Omni-Path, RoCE).
- No specific Linux distribution is required.
- The client is a patchless kernel module; the server components are user-space daemons.
- Runs on top of standard Linux file systems (XFS, ext4) — no dedicated partitions needed.
- Clients and servers can run on the same machine.
- Supports Linux on x86, x86_64, OpenPOWER and ARM64.
- Ships with graphical cluster installation tools.
Performance and Scalability
BeeGFS delivers maximum performance across distributed file content and metadata, requires no kernel patches and is designed to handle very small and very large files alike. It scales transparently from small clusters to enterprise systems with thousands of nodes.
The metadata service manages stripe information for files; file open/close stays out of the data path, so clients reach full bandwidth. BeeGFS fits converged, hyper-converged and classic HPC infrastructures.
Capabilities
High Availability
High-availability storage and metadata mirroring in BeeGFS is delivered via buddy groups. A buddy group is a pair of targets that handles data replication between them — even if half the system fails, data stays accessible.
Built-in BeeGFS Buddy Mirroring:
- Replicates data automatically.
- Handles server failures transparently.
- Provides automatic resync.
Quota Enforcement
BeeGFS lets you define system-wide disk space and file count quotas per user or group.
Two layers:
- Quota Tracking — queries the amount of data and number of files currently used on the system.
- Quota Enforcement — defines and enforces system-wide quota limits.
Access Control Lists (ACLs)
Storage Pools
The Storage Pools feature exposes flash drives (SSD/NVMe) directly to users. Users can move projects onto the flash tier with the beegfs-ctl command-line tool.
- Administrators can group targets into pools — for example, a fast SSD pool and a slower spinning-disk pool.
- Pools can be given descriptive names so users can clearly see where their data lives.
BeeOND — BeeGFS On Demand (Burst Buffering)
Compute nodes are often equipped with internal flash drives, but local storage isn’t suitable for distributed applications on its own.
BeeOND (“beyond” + “BeeGFS On Demand”) instantly stands up a fast, shared, all-flash parallel file system across the assigned compute nodes for the duration of a job. It exists only on the allocated nodes while the job runs and is torn down automatically afterwards.
Typical use cases:
- Burst buffer for HPC workloads
- Scratch / temporary working space
- Dataset cache for ML training
Use Cases
CFD simulation, genomic data analysis, seismic processing, deep learning datasets and large-scale data warehouse workloads.
Licensing & Support
Mevasis is a BeeGFS solution partner in Turkey. For licensing, deployment and operations, contact us.