exponential ramp

Taking the characteristic of the variant-length of vector data records into account,HCSDP algorithm divides uniformly the huge volume of spatial data set into multiple parts and putting them onto the different processing nodes for avoiding the data skew.

在充分考虑空间信息的海量特征以及矢量数据存储记录的不定长等特点的前提下,该算法可实现并行空间数据库中海量空间数据记录在多个存储设备上的均衡划分,以避免出现数据倾斜现象,从而提高了空间数据的检索与查询效率。

Hybrid range partitioning strategy [1] introduces a formula to compute the amount of nodes to distribute data and the data partitioning strategy based on identical range sizes; then, an enhanced hybrid range partitioning strategy [2] achieves data storage balancing and solves data skew between the nodes in the parallel real-time database system by varying range sizes.

混合范围划分方法[1]给出了计算数据分置节点数的公式以及数据划分的方法;加强的混合范围划分方法[2]通过引入可变范围的数据分块,达到了节点间数据存储量的一致,解决了混合范围划分方法的数据倾斜问题。