Disk arrays, coupled with emerging small disk technology, promise to provide a badly needed increase in the performance of secondary storage systems. Because high failure rates arise with a large number of disks, however, simple redundancy schemes are used to ensure data reliability. This monograph investigates the data encoding, performance, and reliability of redundant disk arrays. Gibson reviews the performance advantages of striping data across multiple disks, evaluates the performance lost to the maintenance of redundant data, provides evidence that disk lifetimes can be modeled as exponential random variables, and develops and applies analytic models of data reliability in redundant disk arrays suffering dependent failure modes and featuring on-line spare disks.
