All things being equal with appropriate backups being maintained/performed; then the actual questions should revolve around how much is the cost of downtime? To me it is a much better bargain to have paid for another drive for dual parity as opposed to trying to save a few dollars at the beginning.
Agreed; but how often do you see people listing their backup media budget in the system specs, and then dithering over one extra drive for raidz2?
Everyone needs backup of all data they don't want to lose, regardless of the level of redundancy they have in their pool.
Yep, and starting with raidz1 and a single drive to backup critical data really focuses the mind, doesn't it? It separates the availability from the durability. It leads the user to decide that even more of the pool should be backed up.
The issue is how long it takes to resilver and the likelihood of encountering an error during that time.
How often do your scheduled scrubs register an error, or kill a drive, or otherwise expose a drive that you immediately replace? That is your likelihood. Surfacing those errors before a resilver is the purpose of scheduled scrubs. If your scrubs show a different problem every time, great, raidz2 for sure. Maybe even raidz3!
Restoring from backup is a last resort, and (I claim) sufficiently inconvenient for most FreeNAS users that extra redundancy beyond RAIDZ1 is well worth having, to minimize the chance of having to do so.
I agree. Where I disagree is that anybody should be told this, without first being told to buy a single backup drive before you consider having a second parity drive. That in most cases, the second parity is unimportant next to the question of whether you have a full backup. When your pool is ruined by whatever means, who cares what the parity level was? Let's talk about your backup.
strike, as below: After all, this build doesn't even have ECC. ;)
