Abstract: Connection query optimization technique is very important to improve database performance. This paper proposes an improved connection query algorithm, which combines the Wander Join query algorithm and the ant colony genetic hybrid algorithm to optimize the connection order. After executing the new connection plan, pruning strategy is used to decrease the complexity of sample connection, thus achieving the purpose of reducing storage cost. Theoretical analysis and comparative experiment on TPC-H data set and TPC-DS data set are carried out. Experimental results prove that, under the condition that the sample confidence interval of multi-table connection is greater than or equal to 95%, the connection query algorithm combing the ant colony genetic hybrid algorithm and pruning strategy can reduce the relative error rate by 20% to 70% in comparison to the Wander Join query algorithm.

Key words: data management, database, query optimization, connected graph, mixed algorithm