摘要：Queries containing aggregate functions often combine multiple tables through join operations. We call these queries "Aggregate-Join" queries. In parallel processing of such queries, it must be decided which attribute to be used as a partitioning attribute, particularly join attribute or group-by attribute. Based on the partitioning attribute, we discuss three parallel aggregate-join query processing methods, namely Join Partition Method (JPM), Aggregate Partition Method (APM), and Hybrid Partition Method (HPM). The JPM and APM models use the join attribute, and the group-by attribute, respectively, as the partitioning attribute. The HPM model combines the other two methods using a logically hybrid architecture. Our performance results show that the HPM model outperforms the others. In the performance evaluation, we also incorporate the problem of skew, which may occur at the data level, as well as at the processing level
