Stream-Oriented Processing Of SQL Query Plan Generation Technology Research

In recent years, the Internet industry needs to store and process more and more datasince the advent of Web2.0era. As data size increases, the traditional parallel database hasbeen unable to meet current needs and Hadoop is suitable for lagre scale data processing.Against SQL query plan generation method in the distributed database system based onMapReduce, the existing technology uses a rule-based optimization technique and cannotafford in the heterogeneous environment.In able to overcome the above problems, we use the stream-oriented data processingof the SQL query plan generation method. The method uses an efficient query processingengine analysis.The engine includes pre-processing components, analysis components,optimization components and implement components, and it uses streaming mechanism,continuous query processing techniques, SQL query parsing technology, the cost-basedoptimization strategy and allows user to customize partitioning strategy. So it can supporta variety of SQL queries, thus greatly improving efficiency.In order to prove the effectiveness of the technology, we develop the Alovera, andcompare it with HadoopDB. Experiment results show that: Alovera’s average queryperformance is7.21times of the Hive,3.18times of the HadoopDB. It can generate a SQLquery plan more flexible and handle the data more diversity.