SINGLE-LEVEL, MULTI-DIMENSION, HASH-BASED TABLE PARTITIONING

摘要

In a general aspect, a method includes inserting a record in a partitioned fact table of a star schema data mart. Inserting the record can include determining a first hash value from a first dimension value of the record and determining a first set of candidate partitions from the first hash value. The method can further include determining a second hash value from a second dimension value of the record and determining a second set of candidate partitions from the first hash value. The method can further include comparing the first set of candidate partitions with the second set of candidate partitions to determine a common partition and inserting the record into the common partition.

主权项

1. A computer-implemented method for inserting a record in a fact table of a database, the fact table having multiple partitions, the database having multiple dimension tables corresponding with the fact table, the method comprising:
receiving a request to insert the record in the fact table; determining a first hash value from a first value included in the record, the first value corresponding with a first dimension table of the multiple dimension tables; determining, based on the first hash value, a first set of candidate partitions of the multiple partitions; determining a second hash value from a second value included in the record, the second value corresponding with a second dimension table of the multiple dimension tables; determining, based on the second hash value, a second set of candidate partitions of the multiple partitions; comparing the first set of candidate partitions with the second set of candidate partitions to determine a common partition between the first set of candidate partitions and the second set of candidate partitions; and inserting the record into the common partition.