Lots of disdain for RAID in here. RAID is very useful in a media center.
The major appeal is that it gives you a single consolidated volume in which to put your collection of very large files. Humans, like most animals, like to just throw stuff in a pile
Yes a RAID card can fail, every part of your computer can fail, and each part has a failure mode that
CAN RESULT IN DATA LOSS. Your power supply and memory can very easily corrupt your data, the power supply will probably be noticeable, the memory not so much.
However, going back decades, the highest likelihood of failure is always the hard drive. And it is also the one that has the highest chance to cause data loss upon failure. A lot of this has to do with the fact that it has moving parts and they are moving a lot of the time.
This is what RAID is for (aside from RAID 0). It is to protect you against the
most likely failure you will encounter.
If you want a very affordable solution for Windows I would suggest a Dell PERC5i. I bought one off eBay for $30, including the battery backup. When I was done with them I sold it for $35, which also included two 4xSATA breakout cables. So realistically you could spend $45-$50 and have a fully hardware RAID card that can support up to 8 SATA drives. The one downside to the PERC5i is that you are limited to each drive being 2TB. The array can be larger, but the physical drive can only be 2TB max.
Then RAID isn't an option. RAID will provide a fair level of fault tolerance depending on the version, but it is not expandable. To expand a RAID array, you have to completely rebuild it from scratch.
This is dependent on the RAID solution. The ones built-in to Intel and AMD motherboards are not ideal by any means and they do not support this.
However dedicated RAID cards and the filesystem ZFS (MacZFS on OSX) both very much support this. My own personal experience is that the Dell PERC5i could have an additional drive added to the array and the array would rebuild on the fly. The volume was still
COMPLETELY ACCESSABLE while it did so. I started off with a 3x2TB RAID5, expanded to 4x2TB and then again expanded to 5x2TB. I never lost use of the volume nor did I loose any data. 5 drives on RAID5 is the highest I was comfortable with.
I don't know if the PERC5i supported it, but some hardware cards you can also replace all the drives in the array with larger drives and then once they are all replaced the volume can "grow". So you can start out with 3x2TB drives and replace each one (and rebuilding after each replacement) with a 3TB drive. Then once the last 2TB drive is replaced and you have 3x3TB, you can automagically get more storage.
ZFS does not allow you to expand the individual arrays, but the arrays are subsystems of a "pool". If you make a 3x2TB array, you can not ever change the 3 drive part. You can replace them with larger drives so you get something like 3x4TB drives. However because of the "pool" hierarchy, you can add additional arrays to the pool. Thus a single pool (which appears as a volume) could be made up of 3x2TB RAID5 and 3x3TB RAID5 for 4 discs of data and 2 discs of redundancy. What makes ZFS very different from normal hardware RAID is that all the data is processed with checksums, so that you also protect yourself from random bits of data getting flip flopped as it gets handled by the system.
ZFS is really cool, but it has no support in Windows and even OSX support has some stipulations. It features really are the future of file systems though.