多查询并行处理的模型和同步机制

The Model and Synchronization Mechanism for Parallel Processing in Multiple Query

研究数据库多查询并行处理技术中的模型和同步机制问题．讨论了数据库多查询任务和Ｂ＋树的特点，分析了目前已有的几种查询模型，对多查询任务的并行性开发和同步控制机制等关键性问题进行了较详细的讨论．在此基础上，提出了一种基于Ｂ＋树的数据库多查询任务并行处理模型和有利于加大并行度的同步控制机制，并在ＭＩＭＤ－ＳＭ结构上给出了实现多查询任务并行处理的算法设计思想．

This paper is concerned with the model and synchronization mechanism for the parallel processing technique in database multiple query. The specific features of B+ tree and query tasks are discussed. By using the MIMD-SM structure as the background, a B+ tree-based parallel processing model for multiple query task is developed. Based on the model, the development of the parallelism of the multiple query task and the control mechanism of synchronization, especially the measures to realize the synchronization control mechanism under the presupposition of enhancing the parallelism are investigated. Parallel query algorithms suitable for various cases are designed. To improve the parallelism, the safety node and strong search concepts have been introduced and the external tree idea is used, with also the development of parallelism in the nodes taken into consideration.In the reconstruction of the B+ tree, the side branching technique for shortening the locking time on a node is adopted.