So,
the RAID answer to huge data streams is: make the disks take turns
writing, or reading. This is called striping, and it is one of
the most basic RAID functions. Graphically, it looks like this:
And to the
system, it looks like one big drive. Striping provides fast performance,*
but nothing in the way of fault protection; if an HDD has a problem,
tough luck. Data storage is not redundant. This is designated
RAID level 0.
*You might
think that you'd approach double the performance of a single,
unRAIDed disk (i.e., the file that took 10 seconds to read or
write, now takes 5.2), but such is not the case. The best you
get, using a hardware RAID controller, is around a 30 percent
improvement. (I.e., The file now takes 7 seconds. Or maybe 8.)