B^+树索引文件结构的优化设计

The optimization of B^+-tree index file structure

指出当普通B+树的数据块设计为定长时，数据块长度的选择具有一个最优的选取设计过程。给出了在两种数据库系统模型中，对B+树索引文件结构的优化设计：（1）在具有限量内存的数据库系统DBSLAM（DatabaseSystemswithLimitedAmountofMainMemory）模型中，只是缓存B+树中的部分结点；（2）具有极大数量内存的数据库系统DBSVLAM(DatabasesystemswithVeryLargeAmountofMainMemory)模型中，将缓存B+树中的所有结点。对于DBSLAM和DBSVLAM，分别建立了一个时间开销模型和一个时空开销模型，并以此作为系统优化的基础。

We propose an designing procession of optimization for the data block length while it is fixed inconventional B+-Tree. Here we will discuss the optimization under two different models of database systems:Database Systems with Limited Amoun of Main Memory (DBSLAM) and Database Systems with Very LargeAmount of Main Memory (DBSVLAM). In the model of DBSLAM, we storage only a part of nodes of B+-Tree and build a model of using tAne. In the model of DBSVLAM, we storage all of thenodes of B+-Treeand build a model of using time-space. Base on the models the mathematical methods for optimization arealso presented.