Hardened repository - raid type and rebuild time

[KarmaKuma]

Maybe not the perfect forum for this question, but since it concerns our upcoming Repo HW-refresh, I would like to hear your thoughts anyways.

We currently plan 2x (primary + backup) Dell R760xd2 with 28x 12TB NL-SAS and 1x Perc H755 or H965 Controller.

Now, for multiple reasons we are not quite sure wich raid layout to use, either:
- Raid 60 (2x 13 disk r6, 2x global hot-spare)
- Raid 10 (13x r1, 2x global hot-spare)

One of the reasons is raid rebuild time...

After a lot of reading into raid recovey/rebuild, I *think* the consensus seems, that from a certain span size on (and behind a good raid controller, such as the ones above), a raid 6 rebuild does not really take longer than a raid 1 rebuild. The limiting factor are no more the remaining disks and restripe-calculations, but the write throughput of the replacement disk, the same as with raid 1...

Correct?

[karsten123]

and the amount of data on the volume.

we have one 24 disk machine with 16 TB disks in RAID60 with ~ 20 - 10 free space and a rebuild after replacing a disk takes about a week.

[KarmaKuma]

Could you provide some information on the HW specifics like server model plus raid controller in the system and maybe a "guesstimate" on average system load on the machine while said rebuild was running?

[karsten123]

it is a HPE ProLiant XL420 Gen10
24x SAS 7.2k 16TB RAID60
HPE Smart Array P816i-a SR Gen10
load is 247, because it is a Windows ReFS repository for Cloud Connect and for another VBR deployment.
storage usage is most of the time between 80-90%
enough details?

[KarmaKuma]

Enough details, yes
Thanks a lot!