https://github.com/moosefs/moosefs

https://moosefs.com

MooseFS spreads data over a number of commodity servers, which are visible to the user as one resource. For standard file operations MooseFS acts like ordinary Unix-like file system:

* A hierarchical structure – directory tree
* Stores POSIX file attributes – permissions, last access and modification times, etc.
* Supports ACLs
* Supports POSIX and BSD file locks – including support for distributed file locking
* Supports special files – block and character devices, pipes and sockets
* Supports symbolic links – file names pointing to target files, not necessarily on MooseFS
* Supports hard links – different names of files which refer to the same data on MooseFS

Distinctive MooseFS features:

* High reliability – files are stored in several copies on separate servers. The number of copies is a configurable parameter, even per each file
* No Single Point of Failure – all hardware and software components may be redundant
* Parallel data operations – many clients can access many files concurrently
* Capacity can be dynamically expanded by simply adding new servers/disks on the fly
* Retired hardware may be removed on the fly
* Deleted files are retained for a configurable period of time (a file system level "trash bin")
* Coherent, "atomic" snapshots of files, even while the files are being written/accessed
* Access to the file system can be limited based on IP address and/or password (similarly as in NFS)
* Data tiering – supports different storage policies for different files/directories in Storage Classes mechanism
* Per-directory, "project" quotas – configurable per RAW space, usable space, number of inodes with hard and soft quotas support
* Apart from file system storage, MooseFS also provides block storage (mfsbdev)
* Efficient, pure C implementation
* Ethernet support