Distributed Database Multi-join Query Optimization Algorithm

With the development of computer network technology and digital communication technology, the traditional centralized database becomes more difficult in dealing with the query efficiency and storage speed. As the traditional centralized database can not meet people's needs, so the distributed database system comes on. In distributed database system, data is stored and handled in distributed form.The query issue is the key issue in the distributed database, because the data in the distributed database is stored in distributed form and the database structure is very complicated. The key factor which affects query technology is query optimization problem, because the efficiency of query optimization determines the query efficiency. The distributed query optimization is different from the query optimization of the traditional centralized database in the objective of query optimization. It not only takes into account local processing cost, but also considers the cost of long-range communication. As the multi-relations connection operation is the most commonly used in distributed database query and it also costs the largest communication cost, therefore it becomes the most important problem in query optimization. Although many researchers have done a lot of work in this area, the distributed query optimization is still not enough in some places, for example, how to select the most beneficial semi-join execution sequence, how to select the most beneficial semi-connection among many semi-connections, and how to use the character of the distributed database to improve the parallel execution capabilities of algorithm.In this paper, the basic theories of the distributed database system are firstly introduced, such as the development process, the definition, the classification, the composition, the mode structure and the function of the distributed database. Secondly, the technique of the distributed query optimization and the mostly used connection strategy are introduced. Then an algorithm based on a directed non-cyclic graph is proposed. In the algorithm, an optimal semi-join execution sequence is obtained by setting the node degree, estimating the cost, updating the graph and the dynamic parameter table constantly. At the same time, the algorithm implies parallel operations. In the end, a small teaching management system is introduced in order to verify the performance of the algorithm.