Drives in a NAS age at about the same rate between them. If you had multiple drives around the same age or from the same manufacturing batch, there’s a higher chance they fail around the same age.
After one disk in the array fails, you can insert a new drive and rebuild the array, but during the rebuild, all your drives are in heavier use than normal operation. If you only have one disk redundancy, you’re vulnerable until that rebuild is complete.
oh wow, makes sense. It’s a very slim chance, but not zero. but doesn’t a three mirror setup has the same vulterability.
So if the scenario is that we bought two of the same type, use it equally, they’ll die at the same time. This sentance is also true if we up the number.
The calculations necessary to rebuild a failed drive from parity data stored on the other drives means that for the duration of the time that the array is being rebuilt (aka “resilvered”), you’ll have high activity on the other drives. So during that time there’s an increased chance that a drive that was already on the brink of failure is pushed over the edge. If that happens, your data is gone.
Like I said it depends on your risk tolerance. You may not feel like it’s worth it in your situation. I personally only run a raidz1. I accept the risk that entails, just as people who use raidz2 accept the increased risk that entails over raidz3. There’s no limit to the amount of redundancy you can add. The level of redundancy that’s needed is a decision that only you/your organization can make.
“cascading drive failure” the what now? How do drives die in a domino effect?
three locations seem a bit much, but I totally understand it. Safe storage is tedious, huh.
Drives in a NAS age at about the same rate between them. If you had multiple drives around the same age or from the same manufacturing batch, there’s a higher chance they fail around the same age. After one disk in the array fails, you can insert a new drive and rebuild the array, but during the rebuild, all your drives are in heavier use than normal operation. If you only have one disk redundancy, you’re vulnerable until that rebuild is complete.
oh wow, makes sense. It’s a very slim chance, but not zero. but doesn’t a three mirror setup has the same vulterability.
So if the scenario is that we bought two of the same type, use it equally, they’ll die at the same time. This sentance is also true if we up the number.
The calculations necessary to rebuild a failed drive from parity data stored on the other drives means that for the duration of the time that the array is being rebuilt (aka “resilvered”), you’ll have high activity on the other drives. So during that time there’s an increased chance that a drive that was already on the brink of failure is pushed over the edge. If that happens, your data is gone. Like I said it depends on your risk tolerance. You may not feel like it’s worth it in your situation. I personally only run a raidz1. I accept the risk that entails, just as people who use raidz2 accept the increased risk that entails over raidz3. There’s no limit to the amount of redundancy you can add. The level of redundancy that’s needed is a decision that only you/your organization can make.
well, I never tought of data loss prevention in this way. Thanks.