3 ms·

From what I can tell: Synology SHR is just MDADM RAID with some custom logic to slice up the drives so they can setup RAIDs with hard drives with different size

From what I can tell: Synology SHR is just MDADM RAID with some custom logic to slice up the drives so they can setup RAIDs with hard drives with different sizes.

cvubrugier 6y ago
Exactly! Suppose you have four HDDs: two 1 TB HDDs and two 2 TB HDDs. A single 1 TB partition is created on the 1 TB HDDs. Two partitions (1 TB each) are created on the 2 TB HDDs. A first RAID array (what you may call a slice) is created using the first partition of all four drives (4 x 1 TB, RAID5 = 3 TB). A second RAID array is created using the second partition of the 2 TB drives (2 x 1 TB, RAID 1 = 1 TB). Then the two RAID arrays are assembled with LVM (= 4 TB). SHR is basically LVM over MD.
- magicalhippo 6y ago
  Thank you for the explanation, I was wondering how that worked. How does it work if you then later add a fifth 4TB drive? Anyway, I guess in this sense you could do something similar with ZFS, create a bunch of 1TB partitions and then make vdevs (slice) across a set of partitions on each drive. Not sure if it's a good idea or not though. I do know you supposedly lose some performance when handing ZFS a partition rather than the whole drive, but maybe there's something else I can't think of right now.
  - cvubrugier 6y ago
    > How does it work if you then later add a fifth 4TB drive? If you later add a fifth 4 TB drive, two partitions (1 TB each) are created on it. The remaining 2 TB will be unused: a slice with a single element cannot be replicated. The first slice is expanded from "4 x 1 TB, RAID5" to "5 x 1 TB, RAID5", so the logical capacity of that slice is now 4 TB. The second slice is migrated from "2 x 1 TB, RAID1" to "3 x 1 TB, RAID5". MD is able to change the RAID level (aka the "personality") of a RAID array, which is pretty cool. Then the LVM layer is made aware of the increased capacity of the MD layer with `pvresize`. The solution of using "LVM over MD" suffers from its complexity because you have to manage two distinct layers. For instance, you have to run LVM / mdadm commands in the right sequence to replace a failed drive.

From what I can tell: Synology SHR is just MDADM RAID with some custom logic to slice up the drives so they can setup RAIDs with hard drives with different sizes.

cvubrugier 6y ago

Exactly! Suppose you have four HDDs: two 1 TB HDDs and two 2 TB HDDs. A single 1 TB partition is created on the 1 TB HDDs. Two partitions (1 TB each) are created on the 2 TB HDDs. A first RAID array (what you may call a slice) is created using the first partition of all four drives (4 x 1 TB, RAID5 = 3 TB). A second RAID array is created using the second partition of the 2 TB drives (2 x 1 TB, RAID 1 = 1 TB). Then the two RAID arrays are assembled with LVM (= 4 TB). SHR is basically LVM over MD.

magicalhippo 6y ago
Thank you for the explanation, I was wondering how that worked. How does it work if you then later add a fifth 4TB drive? Anyway, I guess in this sense you could do something similar with ZFS, create a bunch of 1TB partitions and then make vdevs (slice) across a set of partitions on each drive. Not sure if it's a good idea or not though. I do know you supposedly lose some performance when handing ZFS a partition rather than the whole drive, but maybe there's something else I can't think of right now.
- cvubrugier 6y ago
  > How does it work if you then later add a fifth 4TB drive? If you later add a fifth 4 TB drive, two partitions (1 TB each) are created on it. The remaining 2 TB will be unused: a slice with a single element cannot be replicated. The first slice is expanded from "4 x 1 TB, RAID5" to "5 x 1 TB, RAID5", so the logical capacity of that slice is now 4 TB. The second slice is migrated from "2 x 1 TB, RAID1" to "3 x 1 TB, RAID5". MD is able to change the RAID level (aka the "personality") of a RAID array, which is pretty cool. Then the LVM layer is made aware of the increased capacity of the MD layer with `pvresize`. The solution of using "LVM over MD" suffers from its complexity because you have to manage two distinct layers. For instance, you have to run LVM / mdadm commands in the right sequence to replace a failed drive.