嵌入式内存数据库的存储和索引算法研究

提出了用于嵌入式内存数据库的更高效的数据存储算法(EHAS)和索引算法(PMCT-tree)。EHAS算法是基于区-段式,结合类可扩散列的思想,将记录以唯一对应的三元组作为区标号、段标号、地址标号来定位存储的算法。PMCT-tree比典型的T树增加了一种多路分支目录(PMC),它是由T树节点中抽取出的部分边缘阈值构成的。测试结果表明,EHAS算法加快了存储响应时间,且在一定条件下其平均查询时间复杂度可达到常数级;PMCT-tree算法在有效性和查询响应时间上性能良好。

Research on Data Organization and Index of EMMDB

This paper proposes the EHAS(quasi-extendible hashing area-segment) and the PMCT-tree algorithms, which are more efficient ones on data organization and index of embedded main-memory database (EMMDB). The EHAS is a storage algorithm which combines quasi-extendible hashing and is based on area-segment method. It locates and stores records with corresponding unique triples, each of which has three parts separately as an area sign, a segment sign and a storage address sign. The PMCT-tree has a priority match catalog (PMC) more than typical T-tree. The PMC is composed of some edges thresholds, which are extracted from T-tree nodes. Experimental results indicate that the EHAS algorithm accelerates storage response time and its average querying time complexity can reach a constant level under certain conditions; and the PMCT-tree algorithm is good on effectiveness and querying response time.