RAID = Redundant Array of Independent Disks.
Therefore RAID could be used as a safe mechanism to avoid loss of data if the correct RAID architecture is set up (Such as RAID 5 for example). We use it all the time in the security industry for surveillance applications so a business does not lose footage when (not if) a drive fails. You could also assign an additional hot-spare drive which is automatically allocated to the RAID hierarchy when one drive fails. The reason you'd want this is in the event two drives fail in quick succession with RAID 5 your entire array is lost. Once the hot-spare is assigned you can still afford to lose a drive while you acquire a replacement for the initial faulty drive (unless another drive fails before the RAID has successfully rebuilt itself with the hot spare).
The aforementioned scenario can happen if temperatures get out of hand all of a sudden, then the potential is there for drives to start falling in quick succession. This is not something that's likely to happen in a consumer environment. If more pragmatism is required, then RAID 6 may be a better option. It all depends on the application. Certain RAID architectures perform better for write-intensive applications for example.
Most storage applications are read-intensive. That is to say data is more frequently read than written. As such, drives and RAID controllers primarily focus on read performance. Given that RAID 5 works successfully for surveillance storage - think 400Mbps+ of continuous writing to the array and the volumes within, there's no reason why it can't handle the requirements for consumer applications which are in most cases menial in comparison.
If a RAID controller fails it is technically possible to restore the array if the controller is replaced with the exact model running the exact firmware. This is possible because the pointers to access the data is stored on the HDDs' DACSTORE. Therefore, successful restoration requires a controller that reads the data using the same algorithm it was written in. Different controllers vary in the way they manage data which is part of the reason why some controllers outperform others. The other reason of course being the hardware (specifications) of the controller itself, such as the available disk cache, the underlying software and probably design efficiency.
Obviously there are better ways to have data backed up but these solutions cost a fortune and are used by businesses mostly (such as redundant storage servers - think EMC² or NetApp).
It is entirely possible however to get a RAID controller for your PC or storage box at home to at least mitigate data loss. It is advisable to use RAID capable drives for such applications. Many businesses make the mistake of ignoring such advice by buying consumer-grade drives and they eventually end up regretting it. RAID drives are slightly more expensive but not ridiculously priced.
Some controllers won't even successfully build an array with consumer-grade drives.
Given that you are unlikely to fork the cash for at least 4x 8TB drives (to run RAID 5) + a RAID controller you're best setting up a tiny NAS if you have extra hardware lying around.
