并行查询中的索引并行化技术

The index parallelization techonology in the parallel query

提出一种新的索引并行化技术 ,利用多处理机来提高建立在B树结构上的索引性能 ,并提出了一种改进的B树结构———FALNB树 ,基于该树结构 ,以大结点B树方式在多个处理机上分布索引数据 .通过动态调整算法快速进行调整 ,取得很好的负载平衡效果 .给出了时间分析模型 .实验结果显示FALNB树能加快查询响应速度 。

This paper presented a new improved index parallelization technology to improve the performance of index constructed on the structure of B tree by using multi processors. An improved structure of B tree called FALNB tree was put forward. On the basis of this tree structure, the index data was distributed among multi processors by using large node B tree method. The unbalance during process can be quickly adjusted by means of a dynamical adjusting algorithm. As a result, a good balance of load was achieved. The paper proposed a time model for analysis. The experimental results confirm that the query performance was improved as the response time was reduced by the FALNB tree, especially for the index of massive data.