July 5, 201115 yr OK, here's what I think could be made possible.... Please share your thoughts. Here's my BoM and setup scenario: - unRAID version 5.0b7, for support of >2.2TB disks - 6x 2TB disk drives - running unRAID on a hypervisor, like ESXi or KVM, ... - on the hypervisor host, use a RAID controller or SW-Raid and create 3x Raid-0 disks, (each raid-0 disk build from 2x 2TB drives, having 4TB total and striped performance) - forward the raid-0 array/drives as raw disks to the unRAID virtual machine. unRAID should therefore see 3 disks, 4TB each. -> wouldn't this be nice This setup would, of course loose a 2TB disk to the larger parity, resulting in a 8TB array, instead of 10TB with bare-metal unRAID. My hope is, that striping of raid-0 on the host could compensate for the performance loss, from unRAID running inside a VM.
July 5, 201115 yr If you can make the RAID drives appear as usable disks to unRAID then I see no reason why it wouldn't work. I have no idea what that would do for a speed boost though. Peter
July 5, 201115 yr Author Thanks for the feedback. I am sure I can do that. At this very moment I am testing a KVM setup with virtio pv-drivers and raw device mapping for stability. Without RAID-0, disks mapped 1 to 1, I see approx 25MB/sec write and 35MB/sec read performance with all green drives. I have a HW-Raid controller sitting on the shelve and will try to test as soon as my other test has finished.
Archived
This topic is now archived and is closed to further replies.