GlusterFS is an open-source network file system started by Anand Babu in 2005. In October 2011, Red Hat acquired Gluster for $138M and continues to support the open-source project. The name was derived from GNU (which means not Unix) and cluster. During that time, there were many proprietary Unix file systems and storage software.
The Gluster team decided to build open-source software that runs on Linux and commodity hardware. At that time, they were a pioneer in the OSS movement and one of the reasons why open source is so popular today, which many projects established and well maintained. One of the big users of GlusterFS is Facebook, at least in 2015. They published some metrics on their use of the product.
- Datasets: GB’s to PB’s with 100’s of millions of files
- FOPs: 10’s of billions per day
- Namespace Volume Size: TB’s to PB’s
- Bricks (storage units): 1000’s
- Clients: 10’s of thousands
Here are some features for GlusterFS.
- Open Source
- Runs on commodity hardware
- Network File System
- For cloud storage and media streaming
- Brick is a storage unit/directory
- Works on most file systems including XFS
- Nodes create storage pools
- Gluster Volume: a collection of servers belonging to the storage pool
- glusterd (daemon) manages the brick process
- Different volumes are supported for scaling, performance, or both
- Volumes: Distributed, replicated volumes, distributed replicated volumes, dispersed volumes, and distributed dispersed volumes
- Replicated Volume: exact copies are maintained on all bricks