You don't want RAID 0. You want RAID 1. RAID 1 is mirroring. In RAID 1, all but one drive can fail, and you're still OK. The performance and size of the resulting array is limited by that of a single drive (yes, you can improve read performance at the cost of potentially not catching errors, and some controllers offer this).
RAID 0 is striping with no redundancy. In this configuration, if any single drive fails, you lose everything. However, it combines the performance and size of all the drives together.
As for drive failures, I've not had one in several years, now, other than drives that were >10 years old when they did finally fail. I guess I'm either lucky or I treat my drives nicely (or both). It probably helps that I don't generally use external enclosures as those tend to get bumped around a lot. I don't think that they're of lower quality (the drives are often not even powered on before they leave the factory, but this applies to all drives). I think they're just subject to more mechanical and electrical abuse.
Of course, I still keep backups. I also tend to keep my drives reasonably cooled. My last "failure in its prime" was a 100GB (IIRC, around that) Maxtor, back when Maxtor was still a separate company. They even grumbled when I asked for a replacement under warranty. I've had decent luck with Seagate, Western Digital, Hitachi (formerly IBM; actually seem to be very reliable if you avoid the "Deathstars"), and Samsung in various applications from desktop to home server to commercially used arcade games (which is probably the worst possible scenario for a hard drive).
ALWAYS keep backups. Drives are so cheap, now, that there's no excuse for not having them. Perform a test restore every now and then to make sure your backup process works.