Two-level coding for error control in magnetic disk storage products

Error-control coding has played a significant role in the design and development of magnetic recording storage products. The trend toward higher densities and data rates presents continuing demands for an ability to operate at a lower signal-to-noise ratio and to tolerate an increased number of correctable errors. Heretofore, the magnetic disk storage products used coding schemes that provided correction of one burst of errors in a record of length ranging from a few bytes of data to a full track on the disk. In this paper, we present a new coding architecture that facilitates correction of multiple-burst errors in each record in a typical disk storage application. This architecture embodies a two-level coding scheme which offers high coding efficiency along with a fast decoding strategy that closely matches the requirements of on-line correction of multiple bursts of errors. The first level has a smaller block delay and provides very fast correction of most of the errors commonly encountered in an average disk file. The second level, on a larger block size, provides reserve capability for correcting additional errors which may be encountered in a device with symptoms of a weaker component or an oncoming failure. The new IBM 3380J and 3380K disk files use a two-level scheme that is designed around the coding structure of the extended Reed-Solomon code. This design and the related encoding and decoding methods and implementation are presented in detail.