Abstract
In the context of a sharp increase in storage volumes and the widespread use of multi-disk storage, new methods of improving reliability are required. The authors consider the conflicting requirements for redundancy and reliability of data storage. The distributed storage model with multidimensional parity control is described which has redundancy lower than at replications and reliability at the level of complex systems which are applied in the reed-Solomon algorithm. Theoretical substantiation of the method and algorithm of multidimensional parity control is carried out, and also estimation of the parameters of reliability and their advantages over existing methods of distributed storage is executed. The technology is implemented in distributed databases.