Until recently I had been using mdadm with four drives, with each pair in a RAID1 configuration. These drives were 2x2TB, and 2x3TB, however as free space was very short, I purchased another 2x3TB drives with the intention of using mdadm and RAID5.
Upon building the RAID5 array, the initial sync was apparently going to take 11000 minutes to complete at a very slow 4500KB/s speed. I was advised that perhaps ZFS was the best way to go.
So, I installed a minimal version of ubuntu 13.04 and created my ZFS RAIDZ pool using all four 3TB drives (they are all Seagate ST3000DM001 drives), then created a dataset and attempted to copy a file to it... At a whopping 350KB/s!
The machine is a HP Microserver N40L with 8GB of RAM.
I have tested the drives with Windows installed on the same machine and get speeds of ~160MB/s per individual drive (same results in ubuntu using hdparm). I then tried FreeNAS on the machine and set up a ZFS RAIDZ pool using the web interface. I was only connected to a 100mb LAN connection at the time, but the file transfer saturated this connection as expected.
ubuntu-zfs was installed from the ppa:zfs-native/stable repository.
I used the following command to create the pool (serials omitted)
I really hope someone can help me out here, as I don't know why performance is so poor. I have seen online other people that have used the same hardware as me without this sort of issue.
sudo zpool create -o ashift=12 datapool raidz scsi-SATA_ST3000DM001 scsi-SATA_ST3000DM001 scsi-SATA_ST3000DM001 scsi-SATA_ST3000DM001