Originally Posted by Mrzev
I find that hard to believe... but yeah reading up on it seems to be true... but I have had drives goes out in some of my systems at work and I was able to rebuild them successfully. So, whats the best way to ensure stability? Raid 10,51,61?
I did a lot of research on it and it makes perfect sense. Consumer drives have roughly a 1 in 10^14 bits read URE rate: http://www.wdc.com/wdproducts/library/SpecSheet/ENG/2879-771434.pdf
Say you have 3 TB drives, three of them in a RAID 5 array. That's about 6 TB that would have to be read for a rebuild. When you have drives that encounter a URE roughly every 12 TB, that's about a 50% chance of failure:
Enterprise NL-SAS/SATA drives (3.5" 7200 RPM) usually have roughly a 1 in 10^15 bits read URE rate: http://www.wdc.com/wdproducts/library/SpecSheet/ENG/2879-771386.pdf
which is a magnitude higher than the consumer drives. In the case above with these drives, "The probability of successfully completing the rebuild is 93.1%". Much better but still a bit risky that the rebuild fails and you end up having to restore from a backup.
Enterprise SAS drives (2.5" 10K & 15K RPM) usually have roughly a 1 in 10^16 bits read URE rate: http://www.seagate.com/www-content/product-content/enterprise-performance-savvio-fam/enterprise-performance-15k-hdd/ent-perf-15k-5/en-us/docs/enterprise-performance-15k-hdd-ds1797-5c-1504us.pdf http://www.seagate.com/www-content/datasheets/pdfs/enterprise-performance-10k-hddDS1785-8C-1607US-en_US.pdf
which is a magnitude higher than the above. Even if you had 24 of the 1.2 TB 10K RPM drives in a RAID 5 array, "The probability of successfully completing the rebuild is 97.5%" (note: these drives are physically smaller and have much less capacity than the 3.5" drives; I think the highest they make is 1.8 TB).
For consumer drives, RAID 10 is highly recommended
but costly as there's a big loss in array capacity. It also has the huge benefit of avoiding parity calculations so it doesn't hit every drive like RAID 5 and RAID 6 do. As far as UREs on RAID 10:
"No, RAID 10 is not parity and is not subject to URE failures like RAID 5 or 6. There are countless threads on this. It is because of this that RAID 10 is in a completely different reliability category than parity RAID.
All RAID types, except for RAID 0, are able to survive URE without incidence while healthy. It is only while degraded that they are at risk of a URE. For parity RAID, it is when they drop to no remaining parity that encountering a URE during resilver causing the array to fail completely.
RAID 10 does not resilver and does not have this risk. When RAID 10 loses a drive, it remirrors, not resilvers. During a remirror a URE goes unnoticed.
RAID 10 is subject to a URE during this process, but it has no reason to halt a remirror. It would just mean that that one block is corrupt. Generally this has no impact. It is the same as your desktop encountering a URE which is does from time to time. Chances are, no one notices. If it is noticed, the impact is limited to a single file, not the entire array."