RAID parameters
RAID parameters are the settings that say how an array lays data out across its disks: the RAID level, the chunk (strip) size, the disk order, the parity layout and the data offset.
An array turns many disks into one volume by following a fixed pattern. The parameters describe that pattern. They are recorded in metadata, for example on each member disk in Linux software RAID, so the array can reassemble itself at start-up.
What are the main RAID parameters?
- Chunk size, also called strip size or stripe depth: how many consecutive blocks go on one disk before the array moves to the next. SNIA uses "stripe size" for something larger: the data in a whole stripe across the disks.
- Disk order: which disk holds the first chunk, which the second, and so on.
- Parity layout or rotation: in RAID 5, which disk holds the parity for each stripe and how that position moves. Linux md, for example, offers left-symmetric, left-asymmetric, right-symmetric and right-asymmetric.
- Data offset: where array data starts on each disk, after any space reserved for metadata.
Why do the parameters matter when an array fails?
The data is only meaningful when read back in the same pattern it was written in. If the metadata is lost, or an array is re-created with different settings, blocks are read in the wrong order and the volume appears corrupt. Recovering a failed array depends on working out the original parameters. See RAID 5 data recovery.
From the lab
Performed
A Dell PowerEdge T110 II server came in with a three-drive RAID 5 array. One drive had dropped off the array a long time ago. The engineers copied the disks sector by sector, identified the RAID parameters, and left that drive out of the reconstruction. The result was a full file tree without errors. Dell PowerEdge RAID 5 case study
From R3's lab records
Performed
All four drives in a QNAP TS-410 showed as unmounted after a power outage. Every drive copied with 0 bad sectors, but the disk order had changed and the metadata was gone. The engineers worked out a usable configuration. 1.47TB came back, with subfolders intact and the root folder names lost. QNAP RAID 5 case study
From R3's lab records
Related terms
Common misconceptions
Can I re-create the array with default settings to get it back?
Only if the defaults match what the array was built with. Re-creating an array writes new metadata. If the chunk size, disk order, layout or offset differ from the original, the data is read from the wrong places, and the original metadata has been overwritten. On Linux software RAID, mdadm also creates a new RAID 5 as a degraded array with a spare and rebuilds onto it, so a re-create with the wrong settings rewrites the data on one of the disks as well.