The BeeGFS Parallel Filesystem
The BeeGFS Parallel Filesystem
BeeGFS is a high-performing, simple to use parallel filesystem for HPC. Similar to Lustre, it is POSIX compliant and highly scalable in capacity as well as throughput. Its scalability is also achieved by the concept of separating meta- and object storage data and distributing the corresponding server daemons with their back-end filesystems among multiple storage nodes.
BeeGFS storage servers run on top of an existing local filesystem using the normal POSIX interface. There is support for different network technologies (including native Infiniband) and dynamic failover between configured network paths in case one of them fails. The BeeGFS server components are running as user space daemons thus eliminating possible problems with kernel versions and/or patches. The client is implemented in kernel space and integrated into the Qlustar kernel.
BeeGFS Flexibility and Scalability
Multiple BeeGFS services can run on the same physical machine, i.e. any combination of client and servers. Thus, BeeGFS can turn a compute rack, in which nodes are equipped with internal drives, into a powerful combined data processing and HPC shared storage unit. This can eliminate the need for external storage resources and may provide a very cost-efficient solution with simplified management.
BeeGFS supports distributed file contents with flexible striping across the storage servers on a by file or by directory base as well as distributed meta-data. It is optimized specifically for HPC use cases and provides excellent client throughput. Superb meta-data performance with linear scalability is achieved through dynamic meta-data namespace partitioning.
Qlustar BeeGFS Stack
View Larger
The Qlustar BeeGFS stack is provided by a collection of native OS packages created by Qlustar developers and ready to use with Qlustar kernels. For each major BeeGFS release, there is a separate set of packages, so admins have the choice, of what version they want to deploy.
Based on these packages, a server and a client OS image module is provided for each supported BeeGFS version. BeeGFS storage nodes may then easily be setup and managed using QluMan.
Qlustar BeeGFS Features
Qlustar BeeGFS uses ZFS as the default backend filesystem. This allows to take advantage of all its data security and protection features as well as transparent data compression. Ext4 and XFS are also supported.
Disk configurations defined in QluMan are readily assignable to nodes running BeeGFS server components. This allows setting up BeeGFS clusters in no time.
With QluMan’s RXEngine and its command editor you have the possibility to define custom commands specifically acting on BeeGFS nodes. These can then be executed conveniently within the QluMan GUI. Furthermore, by defining corresponding user roles, it easily becomes possible to delegate standard maintenance tasks to less experienced admins without the need to hand out full control. Managing and operating your storage nodes can’t become any simpler and more controlled.
BeeGFS is a registered trademark of Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V.