There are a total of four common RAID configurations. These are:
RAID 0 - Striping
+ Writing and reading speed and storage capacity will increase proportionally to the number of drives, because both all drives are being utilized at once.
- Data safety is halved approximately multiplied by the number of drives. Each new drive creates a new point of failure.
RAID 1 - Mirroring
+ Data safety is doubled for every new redundancy layer (HDD pair). More accurately, for every mirrored drive, the data loss chance is very nearly halved. Assuming an instant crash, the data is backed up. However, there is no protection against logical errors in the filesystem.
+ Read speed is nearly doubled, though write speed is nearly halved.
- Halved storage capacity due to redundancy.
- Must have even number of drives.
RAID 10 - Striping and Mirroring
+ Speed boost and redundancy. Failure chance is about half the chance of a single drive failure. Speed increase is more pronounced in reads than writes.
- Data storage is halved.
- Must have even number of drives, and a minimum of 4.
RAID 5 - Striping and Parity
+ You only lose one drive worth of storage.
- You have a minimum of 3 drives necessary. Redundancy covers one drive failure only.
There are also levels 01, 3, 4, 6, 50... These are mostly upgraded basic levels or (re)combinations thereof.
So, let's assume 4 2 TB drives are present in the system. You'd have:
RAID 0: 8 TB max - Highest Performance (minimum 2 drives)
RAID 1: 4 TB max - Lowest Performance (Minimum 2 drives, must be even drive quantity)
RAID 10: 4 TB max - Highest Performance with redundancy (Minimum 4 drives, must be even #)
RAID 5: 6 TB max - Lowest cost to redundancy (minimum 3 drives)
Just remember, RAID is a bandage and only helps you if you have an instant failure. Degradation isn't accounted for with any of these RAID level, so that slowly failing drive can still bork your system.