3 ms·

> RAID 5 Filesystem noob here. I have a question. Wasn't RAID 5 considered less than ideal because in the event of one disk failing, you're likely to get a fa

• DaSHacka 2y ago
  Yes, this is why RAID5 isn't advised for (most) modern arrays anymore. The time it takes to rebuild a pool full of 10TB+ disks is going to be extremely long, and the chance of a second disk failing during that rebuild is much higher than smaller capacities like 2-3TB that would have been more common 10-15 years ago. For thia reason, its often recommended to use RAID6 for new arrays and treat the second failable disk as strictly a spare for during a rebuild.
• rekabis 2y ago
  Correct. Plus, when rebuilding a failed RAID-5 disk, the system is engaging in mass calculations to rebuild that data from the “fingerprints” stored on the other drives. Conversely, with RAID-10, we have mirrored drives (the 1 part of the RAID) which are then striped across the pairs that exist (the 0 part of the RAID). As such, IIRC rebuilds are a lot faster because it is a direct copy instead of a CPU-intensive rebuild of the data. Plus, you have (technically) a n-drive failure capability, where n = the number of mirrored pairs of drives. Provided that you don’t have more than one drive per pair fail, of course. So for a RAID-10 of four drives, where you have a stripe across two mirrored pairs of drives, you could have up to two drives fail without experiencing data loss, so long as each failed drive was from a different mirrored pair. Plus, RAID-10 is much faster than RAID-5, and much, much faster than RAID-6.