BeeGFS is an open source parallel file system that is suitable for large data storage and environments that require high scalability. Originally known as the Fraunhofer Gesellschaft File System (FhGFS), BeeGFS was designed as a storage solution for high performance computing by the Fraunhofer Institute for Industrial Mathematics in Germany.
As open source software, BeeGFS can be acquired for free. The BeeGFS client is available under the GPLv2 license and the server components are published under the BeeGFS EULA. A support contract for BeeGFS can also be purchased from ThinkParQ which provides support for the software as well as enables additional features for enterprise implementations such as high availability, quota enforcement, and access control lists.
The BeeGFS parallel file system is comprised of four main parts: the management service, the metadata service, the storage service, and the client service.
The management service is a lightweight service that keeps track of the other services and their states. The management service does not store any user data, and typically does not require a dedicated machine.
The metadata service is responsible for storing metadata about the data that is being stored such as directory and ownership information. Data in BeeGFS is stored in chunks using striping that is determined by the metadata service. The metadata service is responsible for providing the stripe pattern to the client, but otherwise is not involved in data access to improve efficiency. The stored metadata is intended to be small and to scale linearly with stored files.
The storage service is the service that is responsible for storing striped user files, referred to as data chunk files, across one or many storage targets. BeeGFS utilizes a scale out design that allows for both capacity and performance to increase as needed. The storage service works with any local Linux POSIX file system. Storage targets typically consist of hardware RAID6 or zfs RAIDz2 volumes.
The final service of BeeGFS, the client service, is responsible for mounting the file system to access stored data. When the client kernel is built, it automatically matches the current running Linux kernel. This eliminates the need for manual updates when the Linux kernel is updated. BeeGFS can also be accessed via NFS, CIFS, or Hadoop. It has been stated that future versions of BeeGFS will include a native client for Windows as well.
BeeGFS also provides advanced features such as Buddy Monitoring synchronous replication to mitigate the risk of hardware failures, storage pools to optimize efficiency of storage resources, and BeeOND (BeeGFS On Demand) to enable the use the internal flash drives of compute nodes as temporary file systems. BeeGFS also features an optional graphical interface for administration and management called “admon”.
BeeGFS Parallel File System Product Brief includes:
Download now to read the full product brief!