I've been trying to find a solution to achieve the following objective: minimum delay redundant network storage for virtualized server and think gluster might be what I need after throwing out options like Lustre, dmraid on Openfiler etc. The configuration in mind is currently this. Application Servers -> Runs a few VM guest OS -> Runs gluster client/server -> VM machine images then stored on mirrored gluster volumes. There will be two storage servers with physical RAID. The concept is that 1. physical RAID 1 catches single disk failure on the storage server 2. gluster mirror on the application server catches single machine failure of the storage servers 3. Avoids any problems caused by a heartbeat/failover delay. 4. If an application server die, the VM images are still on the gluster volumes, I can simply distribute the downed VMs to the other running application server by loading the images. Would this work or am I missing something?