基于多核的嵌入式数据库查询优化-软件工程专业论文.docxVIP

摘 要 随着嵌入式系统应用越来越广泛，系统中产生的大量数据需要被处理，嵌入 式实时数据库作为一个重要的信息存储管理工具，其重要性越来越凸显。目前的 嵌入式数据库中，查询操作流程是比较繁杂的，每一次单独的查询操作都会涉及 到对数据文件的读取。虽然现有的内存数据库避免了大量的 I/O 操作，但是在处理 大量繁杂的数据时，如果有多用户同时操作，不可避免的也会出现频繁的内存置 换，如何实现嵌入式数据库的快速查询，也成为了目前重点的研究问题。 本文主要针对目前国产化嵌入式数据库系统 CS-DB 的查询操作进行了改进。 首先分析了嵌入式数据库的查询操作流程，在此基础上提出了查询操作在多核多 线程下执行的机制。然后针对多核环境下多线程的并发操作，我们提出了多用户 同时操作的优化机制。最后根据查询的对象不同，为数据库制定了新的哈希索引 机制。其间主要涉及到多线程间的调度策略的研究，多线程间同步机制，以及哈 希索引的实现方式等技术。 最后在完成系统的设计后，在 VxWorks 系统平台下进行多用户查询的模拟实 验，通过与原有数据库系统的比较，发现在数据量庞大和在多用户并发操作的情 况下，改进后的数据库比原有数据库在时间上有了明显的速度提升。 关键字：嵌入式数据库 查询操作 多线程 哈希索引 Abstract As the application of the embedded system become more and more widespread, there is a large amount of data which is produced by system will need to be processed. As an important management tools of information storage, embedded real-time database’s importance become more and more obvious. In current embedded database, the process of query operation is very complicated. Each single query operations involve the read of data file. Although a large number of I/O operations are avoided by the existing memory database, in dealing with a large number of multifarious data or multi-user concurrent operation, it is inevitable that frequent replacement in memory. How to realize high speed query for embedded database have become the focus of the present research problem. This paper is mainly aiming at improving the query operation in the embedded database system of CS - DB. First of all, analysis of the query operation process in embedded database, and based on this, raising the execute mechanism of the query operation within the multi-core architecture and multi-thread. Then for simultaneous operation of multi-thread under multi-core architecture environment, we put forward the optimization mechanism of multi-user concurrent operation. Finally, according to the diversification of query objects, we established hash index mechanism for database. The optimization design main involves the scheduling strategy within multi-thread, multith