Depending on where you look up the definition you will find either, Redundant Array of Inexpensive Disks or Redundant Array of Independent Disks. I prefer the second as the disks used in an array can be anything but inexpensive!
Why RAID? The biggest bottleneck in any computer is the transfer of data from a hard drive to volatile RAM. Once you click that power button what is happening while you wait for that desktop to appear is the transfer of data from that hard drive, when you open an application, data is transferd, so we wait. So bottom line anything that can be done to increase how fast that information is moved will overall and significantly improve the speed of any computer.
Another issue with hard drives is that they crash. The question is not IF a hard drive will crash the question is most certainly it will but what is unknown is WHEN. When this does occur the data on the drive may be partially or completely lost. This is another benefit of RAID as it allows data to be mirrored over several disks and thus decreasing the likelihood of data loss.
There are two basic forms of RAID 0 and RAID 1, other levels of RAID incorporate forms of the first two.
This form of RAID will distribute or interleave data over several disks. This will increase performance. The disadvantage of this form of RAID is the increased probability of data loss. Simply striping two drives doubles the odds of data loss. A simple analogy is if you were to rip every other page out of a book would make the book virtually incomprehensible. The same is true if one of the disks in a RAID 0 array fails or crashes total data loss occurs. It is important to note that this type of array is recommended to store the OS and application installs or recoverable data. This data also greatly benefits from increased performance. So this is perfect for applications where performance is utmost and storage will be of non volatile data or recoverable data.
This form of RAID will duplicate (mirror) data across other disk(s) in the array. The advantage here is that data integrity is maintained in the event of a disk crash or failure. This provides no improvement in performance but adds the security of maintaining or reducing that chance of loosing mission critical or un recoverable data or to maintain zero down time, as one drive fails the mirror kicks in to take over.
Any RAID array can be configured to be internal, contained in the same case as the server/PC or built into an external enclosure. There are advantages and disadvantages to both, the situation can / may depend on your setup. Important to note that external solution does NOT necessarily compromise on speed as interface range from low end USB 2.0 @ 480Mbps, Firewire 800 @ 800Mbps and eSATA @ 3Gbps (3000Mbps). External solutions are often afterthoughts or add on for additional storage in an already established network or setup.