Abstract: An improved storage device, such as a disk drive, for use in an array of drives. Each drive has logic and Exclusive Or calculation means. Data to be written is sent to the target disk drive from the host computer under a new command that instructs the target drive to generate part of the new parity, and may also identify the parity drive. The target device reads the old data from it's media, generates a parity syndrome, stores the new data and then initiates communication with the identified parity drive, instructing the parity drive to read the old parity and generate the new parity information from the parity syndrome and the old parity. The parity drive, after generating the new parity, writes the new parity onto it's media, all without further host or host I/O action.

Abstract: A multiple transducer disk file format and head addressing/switching control circuitry is shown and described which permits reading all sector identifier (ID) fields during a single rotation of the media using uniformly staggered sectors. This results in an average access latency of one-half revolution of the media and enables the use of maximum or selectively reduced data rates secured by successively storing data in sectors separated by sector fractions which are multiples of the uniform staggering between sectors within a cylinder.

Abstract: A method and apparatus for performing data base searches in which the host processor and main memory are free for other processing tasks between the time that the host processor requests the search until the search results are reported back to the host processor. To commence the search, an input/output controller communicates from the host processor to a record scan circuit values of a skip length, a key length and a data length. While data records are received serially from disk files, within each data record, a length of data equal to the specified skip length is initially skipped. Following this, a search argument is compared with a length of data specified by the key length value. This comparison operation is alternated with skipping of data specified by the data length value until the end of the record is reached or until a specified number of comparisons has taken place. The data record is stored as it is received from the files.