Gluster

Gluster is a distributed file system that supports both replicated and dispersed data.

Supporting dispersed data is a differentiating feature. Only a few can distribute the data in a erasure-coded or RAID-like fashion, making efficient use of space while providing redundancy. Have 5 cluster members? Just add one ‘parity bit’ for just a 20% overhead and you can loose a host. Add more parity if you like with the incremental cost. Other systems require you to duplicate your data for a 50% hit.

It’s also generally perceived as less complex than competitors like Ceph, as it has fewer moving parts and is focused on block storage. And since it uses native filesystems, you can always access your data directly. Redhat has ceased it’s corporate sponsorship, but the project is still quite active.

So you just need file storage and you have a lot of data, use gluster.