The FUSE system was originally part of AVFS (A Virtual Filesystem), a filesystem implementation heavily influenced by the translator concept of the GNU Hurd. It superseded Linux Userland Filesystem, and provided a translational interface using lufis in libfuse1.
FUSE was merged into the mainstream Linux kernel tree in kernel version 2.6.14.
The userspace side of FUSE, the libfuse library, generally followed the pace of Linux kernel development while maintaining "best effort" compatibility with BSD descendants. This is possible because the kernel FUSE reports its own "feature levels", or versions. The exception is the FUSE fork for macOS, OSXFUSE, which has too many differences for sharing a library. A break in libfuse history is libfuse3, which includes some incompatible improvements in the interface and performance, compared to the older libfuse2 now under maintenance mode.
As the kernel-userspace protocol of FUSE is versioned and public, a programmer can choose to use a different piece of code in place of libfuse and still communicate with the kernel's FUSE facilities. On the other hand, libfuse and its many ports provide a portable high-level interface that may be implemented on a system without a "FUSE" facility.
Operation and usage
A flow-chart diagram showing how FUSE works: Request from userspace to list files (ls -l /tmp/fuse) gets redirected by the Kernel through VFS to FUSE. FUSE then executes the registered handler program (./hello) and passes it the request (ls -l /tmp/fuse). The handler program returns a response back to FUSE which is then redirected to the userspace program that originally made the request.
To implement a new file system, a handler program linked to the supplied libfuse library needs to be written. The main purpose of this program is to specify how the file system is to respond to read/write/stat requests. The program is also used to mount the new file system. At the time the file system is mounted, the handler is registered with the kernel. If a user now issues read/write/stat requests for this newly mounted file system, the kernel forwards these IO-requests to the handler and then sends the handler's response back to the user.
Unmounting a FUSE-based file system with the fusermount command
FUSE is particularly useful for writing virtual file systems. Unlike traditional file systems that essentially work with data on mass storage, virtual filesystems don't actually store data themselves. They act as a view or translation of an existing file system or storage device.
In principle, any resource available to a FUSE implementation can be exported as a file system.
On-disk file systems
Conventional on-disk file systems can be implemented in user space with FUSE, e.g. for compatibility or licensing reasons.
Linear Tape File System: Allows files stored on magnetic tape to be accessed in a similar fashion to those on disk or removable flash drives.
retro-fuse: retro-fuse is a user-space filesystem that provides a way to mount filesystems created by ancient Unix systems on modern OSes. The current version of retro-fuse supports mounting filesystems created by Fifth, Sixth and Seventh Edition of Research Unix from BTL, as well as 2.9BSD and 2.11BSD based systems.
Layering file systems
FUSE filesystems can create a view of an underlying file system, transforming the files in some way.
Borg (backup software): Deduplicating backup program that allows backup archives to be mounted as FUSE filesystems.
Restic: Free, fast, efficient and secure backup software uses FUSE to be able to browse all of your backup snapshots as a regular file system
SPFS A file system for Spectrum Protect, designed to mount the backup server filespace anywhere on your server, and use the features included from the backup server (encryption, de-duplication, compression, filtering etc). This is a WORM file system.
Remote/distributed file system clients
CernVM-FS: A distributed read-only software distribution system, implemented as a POSIX filesystem in user space (FUSE) using HTTP transport, to deliver software in a fast and reliable fashion at global scale.
CloudStore (formerly, Kosmos filesystem): By mounting via FUSE, existing Linux utilities can interact with CloudStore
ExpanDrive: A commercial filesystem implementing SFTP/FTP/S3/Swift using FUSE
Lustre Cluster filesystem will use FUSE to allow it to run in userspace, so that a FreeBSD port is possible. However, the ZFS-Linux port of Lustre will be running ZFS's DMU (Data Management Unit) in userspace.
MinFS: MinFS is a fuse driver for Amazon S3 compatible object storage server. MinFS lets you mount a remote bucket (from a S3 compatible object store), as if it were a local directory.
MooseFS: An open source distributed fault-tolerant file system available on every OS with FUSE implementation (Linux, FreeBSD, NetBSD, OpenSolaris, OS X), able to store petabytes of data spread over several servers visible as one resource.
Nexfs: A commercial Linux file system that combines Block, File, and S3 compatible Cloud & Object storage into a single pool of POSIX compatible storage.
ObjectiveFS: Distributed filesystem with object store backend (Amazon S3, Google Cloud Storage or S3-compatible object store) using FUSE
Rclone can mount a variety of remote / cloud storage with FUSE.
s3fs: Gives the ability to mount an S3 bucket as if it were a local file system.
Sector File System: Sector is a distributed file system designed for large amount of commodity computers. Sector uses FUSE to provide a mountable local file system interface.
SSHFS: Provides access to a remote filesystem through SSH.
Transmit: A commercial FTP client that also adds the ability to mount WebDAV, SFTP, FTP and Amazon S3 servers as disks in Finder, via MacFUSE.
WikipediaFS: View and edit Wikipedia articles as if they were real files
Wuala: Was a multi-platform, Java-based fully OS integrated distributed file system. Using FUSE, MacFUSE and CBFS Connect respectively for file system integration, in addition to a Java-based app accessible from any Java-enabled web browser (service discontinued in 2015).
IndexFS: A remote file aggregating filesystem with transparent CURL access to distributed files.