Unfortunately it is the other way around. Lets assume the probability of a single drive failure over a period of 5y is 10%. This means, if you have an array made of 10 drives the probability of a failure within this array within 5y is 100% (vs 20% for R1 with 2 drives).Brah wrote:@X9200 - Thanks for the continued suggestions. Any issue I may have with RAID is a) any assumption that it is an alternative to backup for disc failure, and b) for a 2x config, that it is not much safer than 1x, and less safe than 1x with a backup.
[..]
Why I regard RAID >1 higher that RAID 1 is that that involves more drives = less likelihood of total loss (save for things like fire, flood, theft, etc. ). Like you said, you could have RAID 1 with >2x but not sure why anyone would do that.
Besides, for R1 you have always 2 ready copies. If one fails, you replace the drive and your protection level is immediately restored. If 1 drive fails within an R5 array and you replace this drive with a good one it still takes hours to rebuild the array. If another failure happens during the rebuilding, and this is really very intensive drive work time, then all the data is or could be gone. This is a pretty serious threat as this is linked to something know as unrecoverable reading error (URE) that is said to possibly happen roughly once per 10-15TB of data read from the disk. This likelihood is defined in the harddrive spec. and I personally think it is a bit overdone but nevertheless it is something real. It never happened to me and I've been using R5s for 7 years at this point rebuilding the arrays a few time but still.
One important clarification: I am recommending RAIDs, as an n-copy of the data and never as a single archive unit. I assume that you will have at least:
1. original data on the original storage
2. a copy of the above on a RAID
so under normal, system healthy circumstances you will have at least 2 copies of the data of which one is RAIDed.