A~*Algorithm Based Research And Implementation Of Grouped Multiple Query Optimization

Multiple Query Optimization(MQO)is an efficient way to accelerate the processing of high-concurrency queries.Traditional MQO focus on scenarios of multiple queries with strong similarity.Algorithms may adopt heuristic search strategy to speed up the process of getting suboptimal global plan.The A*based MQO algorithm is the representative of these traditional algorithms and many improvement strategies were proposed to improve it.However,traditional MQO algorithm can not deal with the scenario of multiple queries with low similarity.In such cases,search space is huge but few parts among queries can be shared,which may lead to a bigger overhead than benefit.So in order to improve the efficiency of traditional MQO algorithm in scenario of multiple queries with low similarity,an A*algorithm based grouped MQO algorithm is proposed in this paper.Firstly,a grouping strategy is proposed to speed up the process of the A*algorithm based MQO algorithm.After that,by analyzing how Impala processes SQL queries,we modify the procedure of the system to support MQO with the integration of the proposed algorithm.Finally,our work is evaluated by TPC-DS from two aspects:the performance of MQO algorithm and the performance of grouping strategy.Experimental results show that the proposed MQO method in this paper can effectively improve the performance.