I’m looking for 16TB HDDs. They’ll be for fairly light usage. Immich will be the heaviest thing running on it.
New? Used? Certified? Like this?
I’m looking for 16TB HDDs. They’ll be for fairly light usage. Immich will be the heaviest thing running on it.
New? Used? Certified? Like this?
Sounds a bit like not enough redundancy. Once you go into redundant mode, the individual disk quality is no longer nearly as important. 2 or 3 disk redundancy, and you can use whatever garbage comes your way.
That assumes you don’t value your time spent dealing with troubles that come.
Like the other person said, it’s fine if you don’t, but for me it’s worth a little upfront cost to have to deal with less ordering new drives, putting the drive in the server, monitor rebuilding of the array, ect…
None of that is an excuse for lack of proper backups. Because even new drives can fail catastrophically.
I don’t understand how this follows from what I said. 🤔 I called for increasing redundancy to compensate for the increased risk of failure. That’s the purpose of redundancy. Reducing the time spent dealing with troubles. Unless you consider replacing a disk to be a significant time spent. To me it isn’t because it’s fairly trivial in my setup. Perhaps it’s more work in other setups.
Depending on the prices, you may even be able to add significantly more redundancy by using recertified disks, potentially reducing the risk even more than running new drives. E.g. 4-disk redundancy vs 2-disk for the same price. Running a significantly more redundant setup not only decreases the probability of an array failure but it should also reduce the mechanical load each disk experiences over time which should further decrease failure risk.
All well and good until you lose another disk 2 days into re-striping. Which is not that uncommon because that puts a lot of load on the surviving disks! Remember, RAID is not a backup.
That’s why the extra redundancy. The probability of 2 or 3 disks failing should be significantly lower than 1 disk failing. I currently run 2-disk redundancy. If 1 disk fails, I’d replace it. If a second disk fails while the replacement is being resilvered, I’d shit a brick, stop the resilver and make an incremental backup to ensure I won’t lose data if another disk fails due to the resilver load. Then I’d proceed with the resilver. RAID is not backup and the extra redundancy is there to reduce the probability to have to spend time restoring backups. Increased redundancy can compensate for individual disk reliability.