Question: I’m not looking for what Synology Hybrid RAID (SHR) allows the user to do, but rather, what is the concept that SHR uses to allow you combine disks of different sizes? In most RAID implementations such as ZFS, Linux RAID/mdadm or Btrfs, only disks of the same size can be combined. I have also heard that SHR is based on Linux RAID, and if so, how can it allow me to use different sized disks when standard Linux RAID can’t? Does it use a scheme where disks are partitioned into equal sized chunks and then setup in an array? The documentation on Synology’s website explains the outcome that SHR can achieve, but now how it achieves it, which is what I would like to know.
Answer: Synology Hybrid RAID works as follows:
- One md RAID 1 across all disks for the NAS root file system (about 2.3GB on my system).
- One md RAID 1 across all disks for swap (about 2GB on my system).
- One LVM volume group containing multiple md RAIDs (either RAID 5/6 or RAID 1 for two devices).
If you have 2?1TB and 2?2TB, there would be two md devices in the VG: One with 1TB (minus root and swap) across all devices and another one with 1TB across the two larger devices.
When you replace one of the 1TB devices with a 3TB device, the first md will be repaired, the second md will be reshaped to a RAID 5 containing the “second TB” of each of the larger 3 devices and the last TB is not available.
As soon as you replace one of the 2TB devices with another 3TB device, a RAID1 of the last TB of the now two 3TB devices will be created and added to the LVM VG.