Parallel Computation Method for Fragmentation of Distributed Database Data Based on Rank-Based Approach

An analysis of problem of the fragmentation of distributed database data showed that the formal model is the so-called integer linear programming problem with Boolean variables. Unfortunately, the task of the ILP with the BP belongs to the class NP (non-deterministic polynomial), which can easily be solved even with the use of modern computers. The proposed method of parallel computing for fragmentation of distributed database data is based on the ranking approach to its solutions, the principles of optimization in the direction and strategies, for cutting off non-prospective solutions. The method allows us to determine the local extrema at the vertices of the graph and determine the global extremum on their basis.