Global file system


In computer science, global file system has historically referred to a distributed virtual name space built on a set of local file systems to provide transparent access to multiple, potentially distributed, systems. These global file systems had the same properties such as blocking interface, no buffering etc. but guaranteed that the same path name corresponds to the same object on all computers deploying the filesystem. Also called distributed file systems these file systems rely on redirection to distributed systems, therefore latency and scalability can affect file access depending on where the target systems reside.

History

The Andrew File System attempted to solve this for a campus environment using caching and a weak consistency model to achieve local access to remote files.
More recently, global file systems have emerged that combine cloud or any object storage, versioning and local caching to create a single, unified, globally accessible file system that does not rely on redirection to a storage device but serves files from the local cache while maintaining the single file system and all meta data in the object storage. As described in Google's patents, advantages of these global file systems include the ability to scale with the object storage, use snapshots stored in the object storage for versioning to replace backup, and create a centrally managed consolidated storage repository in the object storage.