pwm Posted November 25, 2017 Share Posted November 25, 2017 On 2/8/2017 at 9:41 PM, jbrodriguez said: But wouldn't the alternative be something like ZFS with realtime checksum integrity (plus periodic scrubs), at the expense of hard-to-expand vdevs ? I guess I'm saying there are pros & cons ... caveat emptor applies Checksumming just tells what is broken. In some situations, you can use the checksumming to react early and repair smaller problems from parity. It also helps with silent errors - when a drive produces wrong content without reporting a read error. But when it comes to broken disks, your only options are enough parity disks to do full rebuilds, and a full backup in case you have too many problematic drives. Remember that a failing power supply can kill every single disk in a RAID - so whatever number of parity drives you have, you need to still consider backup or accept the chance of a potentially huge loss of data. Quote Link to comment
SSD Posted November 25, 2017 Share Posted November 25, 2017 With good backups you can recover from anything. The value of dual parity at that point is really uptime. Because once a second drive drops offline with single parity, the whole array drops offline. That second parity would keep it running. IMO, backups are far more important, at least for data that is unrecoverable via other means. For example, a movie can be repurchased, but pictures and home movies can be irrevocably lost. Quote Link to comment
pwm Posted November 25, 2017 Share Posted November 25, 2017 Yes, RAID is for increased availability while true data safety comes from multiple backups to multiple geographical locations. Quote Link to comment
landS Posted November 25, 2017 Share Posted November 25, 2017 (edited) I moved to Unraid from WHS with ~12 1TB disks after having a dual disc failure result in total array loss. My backups at that time were incremental cd burns, and recovering the data took weeks. . Now our computers are mapped to Unraid as our primary storage (tower 1) which has dual parity and 4 4TB data disks. Documents (work, insurance, pictures) are backed up to crashplan via docker. Media is not sent to crashplan as the physical disc's are all boxed up in my crawl. . The backup server (tower 2) contains a mirror of tower 1...and is comprised of dual parity with 4 4TB data disks (all hgst on this one). . FYI, document folders on my backup server use a copy when modified while media folders use a copy when new system if backup. . Checksum on both servers has actually prompted me to replace a Seagate that otherwise showed as healthy on the primary server... And the backup server allowed me to restore the affected media files while crashplan recovery provided the scrambled family photos recovery. Edited November 25, 2017 by landS Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.