RAID is a system for building large, fast, secure high-speed hard drives from less-expensive equipment.
A RAID array is a Redundant Array of Inexpensive Disks. It's a system for building large, fast, secure high-speed hard drives from smaller, less-expensive hard drive equipment. The EServer's primary storage array consists of four 400GB SATA2 serial ATA hard drives, in an internal RAID 5 configuration. The primary array also has battery backup to cache data being read from and written to the array drives, so that the data is doubly protected against power outages.
The EServer's secondary storage array consists of an external tower with five 250GB serial ATA hard drives, in a configuration known as RAID XL (a variant of RAID level 5 that uses 64- or 32-bit word transfers and a dedicated parity hard drive to transfer data in parallel across multiple drives. That contrasts with traditional techniques of using 64-Kbyte block data transfers across one drive at a time).
RAID protects server data by spreading every byte among multiple disk drives (in our primary array, three drives), then calculating and storing parity information on a fourth drive. This reduces the mechanical limits on the speed of reading or writing data, and the redundancy allows any drive to fail without causing the array itself to lose any data. When a failed drive is replaced, its contents can be reconstructed from the information on the remaining drives in the array.
In fact, a few times, in 2005 and 2006 we actually had to do this. It worked almost exactly as it should have, even once when we had two drives fail, the second soon after the first.
RAID also increases disk subsystem performance by distributing read tasks among several drives, allowing the same data to be retrieved from different locations, depending which happens to be closest at hand when the data is requested.
The EServer RAID secondary (external) array has five discrete serial ATA cables, each to its own discrete SATA channel, yielding a total 7.5 gigahertz, 6.0 gigabit-per-second bandwidth--more than sufficient to manage streaming, file server access, our content management systems and high-speed webserving.