The optimization of databases replication in distributed information systems

New mathematical models of optimal distribution of databases replication in nodes of distributed information systems are formulated by the criteria: minimization of maintenance costs; restricted memory resources; minimizing synchronization time; minimizing the average time needed to search information. Precise solutions of the problems with the use of dynamic programming methods are constructed, Bellman recursive equations are obtained. The general scheme of the computational algorithm using the “greedy” choice procedure is presented and an algorithm for improving the obtained result is proposed. The strategies of greedy choice were investigated, the choice of criteria in the strategy of greedy choice is substantiated. The proposals have been formed regarding the formation of a balance between the accuracy and computational complexity of the algorithm through the introduction of a restricted search strategy. The computational complexity of the algorithm is estimated and its correctness is substantiated.