JOIN algorithm for handling data skew in computer cluster;

On the basis of the analysis of the characteristic of data skew, a hybrid skew hash join algorithm is presented, which can surmount the negative effects of static data skew and motion data skew.

The algorithm can switch on different modules in the case of data skew and eliminate load unbalances caused by data skew.

When excavating with parallel association rules, the two data distribution characters, data skewness and workload balance,will affect the validity of pruning.

The enhancement makes decisions in the local data which belong to the sub-classes of an internal class to lighten the influence of global data distribution,and partially overcomes the problem of data skewness.

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.

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.