Most raid failures are the fault of the user not understanding how raid works and leaving their raid improperly managed.
First, buy good drives. Buying desktop drives that suck are a great way to end up with a corrupt SAN. In general you want drives that support TLER so if something goes wrong, you hear about it right away, rather than the drive trying to fix the problem silently while causing the disk to have long delays. You don't have to go super expensive enterprise, but buying from the bottom of the barrel, or is a 'green' or efficient drive is a great way to lose data.
Second, no matter if you are using hardware or software raid, set up the monitoring utilities properly. You need an immediate alert if a drive is going bad or has failed. SMART should be enabled and when it alerts, replace the drive. The software should also do a verify, patrol read, or scrub (different terms for the same thing) that occasionally check the surface media of the disk and the validity of the data. If anything comes back with an error, you replace the disk right then.
Lastly, one of the big issues why people lose raids all at once is they go out and buy a huge stack of the same kind of disk, made on the same day, with the same firmware, and possibly the same inattentive QA person. Sometimes when problems crop up, it can be a systemic problem with that model, and when all you have is one model, bye bye everything.