Group-by processing for data containing singleton groups
View Patent ↗According to one embodiment of the present invention, a system performs a grouping operation for a database query. The system assigns data elements to groups and aggregates information for a group in response to assigning the group two or more data elements. The system passes the aggregated information for a group of two or more data elements for processing in accordance with the query, and passes information for a data element of a single-member group in a received form for processing in accordance with the query. Embodiments of the present invention further include a method and computer program product for grouping data elements in substantially the same manners described above.
1. A method of performing a GROUP BY operation for a database query comprising:
receiving one or more blocks of data elements, wherein each data element comprises compressed information;
assigning the data elements to groups to form one or more single-member groups and one or more plural-member groups, wherein assigning the data elements comprises:
assigning a data element to a new group to form a single-member group and storing a pointer to a received block containing that data element; and
assigning a data element to an existing group to form a plural-member group, decompressing compressed information of the data elements assigned to the plural-member group, and forming aggregated information for the plural-member group from the decompressed information; and
passing the aggregated information for each plural-member group for query processing; and
passing the compressed information of a data element of a single-member group for query processing.
2. The method of claim 1 , wherein the GROUP BY is specified by a Structured Query Language statement containing a GROUP BY clause.
3. The method of claim 1 , wherein a first block of data elements contains a data element belonging to a plural-member group, and forming aggregated information for the plural-member group comprises forming aggregated information for each group having a data element in the first block of data elements.
4. The method of claim 1 , wherein a first block of data elements comprises a data element belonging to a plural-member group, and the method further comprises:
maintaining a presence indicator of each data element within the first block; and
removing a presence indication for a data element having membership in the plural-member group.
5. The method of claim 4 , wherein passing the compressed information comprises passing presence information and compressed information for each element of the first block.
6. The method of claim 1 , wherein assigning the data elements to groups includes:
applying data of the data elements to an associative array to determine database object elements within a same aggregation bucket based on one or more aggregation keys specified by the query.