混合动态数据库集群的并行空间连接优化算法

传统算法数据划分冗余度和倾斜度高,无用连接数据多,降低负载均衡性,对整体效率产生不好的影响,不适于实际应用。为此,面向混合动态数据库集群提出一种新的并行空间连接优化算法。采用网格划分法对数据进行划分,依据空间数据划分结果获取数据分布状态,计算节点按照数据分布状态得到候选集。通过平面扫描形成若干子空间连接的子任务,利用构建节点花费模型,依据花费模型对并行空间连接所需的平均节点访问个数进行评估,把候选任务集分配至不同计算节点,在不同节点执行并行空间连接操作。通过边界过滤策略,删除不可能有结果的元组,提高效率,增强算法的实用性。实验结果表明,所提算法适于实际应用,效率高。

Research on parallel spatial connection optimization algorithm of hybrid dynamic database cluster

Traditional algorithm data partition redundancy and high degree of slope, useless connection data, reduce load balancing, and have a bad effect on the overall efficiency, which is not suitable for practical application. To this end, a new parallel spatial connection optimization algorithm is proposed for hybrid dynamic database cluster. Meshing method is used to divide the data, and the data distribution state is obtained based on the partition result of spatial data, and the computing node obtains the candidate set according to the data distribution state. The formation of a number of sub space connected sub tasks by plane scanning, construct the node cost model, based on the cost model of average node spatial parallel connection required access number to assess the candidate set of tasks assigned to different computing nodes in different nodes perform parallel spatial join operation. Through the boundary filtering strategy, remove the impossible tuples, improve the efficiency and enhance the practicability of the algorithm. The experimental results show that the proposed algorithm is suitable for practical application and has high efficiency.