It depends on just how well protected your backups need to be. What probability of losing the backups through disc failures can you accept?

No raid means if you lose a disc you have lost any backups it contained pars of. RAID 0 is actually worse than no raid in that respect, since it is probable that each drive contains part of every backup. RAID 10 and RAID 5 provide some redundancy so that you don't lose anything from a single disc failure. The critical question is what is the probablity of a second relevant disc failure before you have completed recovery from the first disc failure? With RAID 5 evey disc in the array is relevant, whereas with RAID 10 only 1 disc in the array is relevant, so the probability of a second relevant failure is higher with RAID 5 than with a RAID 10 array with the same effective capacity. If you do frequent log backups, these will impact the time taken to complete recovery: the imact is worse for RAID 5 than for RAID 10. For both RAID 5 and RAID 10 teh time to complete recovery is shorter if teh array includes a hot spare disc, so that the RAID controller can start recovery as soon as it detects a failure rather than having to wait for human invention to replace the failed drive. Also with both RAID 5 and RAID 10, it can be useful to hold log backups on a separate array from full and incremental backups if there are frequent log backups to eliminate some of the impact of log backups taking place while an array is recovering from disc failure.

Although RAID 10 delivers a clearly lower risk than RAID 5 of completely losing some backups, this may not be significant enough to make using RAID 10 for backup worth the extra cost. An 11 disc RAID 10 - 10 active discs plus 1 hot spare - has the same data capacity as a 7 disc RAID 5 (6 active discs and a hot spare), so if you need that capacity for your backups using RAID 10 will require 4 extra drives and the slots to mount them in - and unless the cost of losing backups is very high indeed it may not be worth paying for that just to reduce the expected frequecy of losing backups perhaps from once in 3 million hours to once in 12 million hours; but of course that depends on the expected cost of losing backups, where one has to take into account the risk of losing the databases that these are backups of. RAID 5 with a hot spare costs an extra 2 discs and their slots compared to no RAID, and that's a cost usually worth paying to reduce the expected frequency of losing some backups perhaps from once per 1700 hours to once per 3 million hours. The repeated "perhaps" in there is because the expected failure frequency depends very much on (a) reliability of RAID controller and (b) reliability of individual drives and these vary greatly from model to model and are changing over time as the state of the art changes so you have to do the calculations using the figures for the equipment types you intend to use, and of course the disc space needed to hold your backups.

edit; some typos