Efficient Execution of Range-Aggregate Queries in Data Warehouse Environments

摘要

Range-aggregate queries on the data cube are powerful tools for analysis in data warehouse environments. Cubetree is a technique materializing a data cube through an R-tree. It provides efficient data accessibility, but involves some drawbacks to traverse all the internal and leaf nodes within given query ranges to compute range-aggregate queries. In this paper, we propose a novel index structure for materializing a data cube, called aggregate cubetree. Each record in all internal nodes of an aggregate cubetree stores the aggregate value of all child nodes of it. Therefore, range-aggregate queries on an aggregate cubetree can be processed without visiting child nodes whose parent node is fully included in the query range, by using the aggregate values in the records of each internal node. The aggregate cubetree is superior to the original cubetree because it can execute queries with a smaller number of node accesses, and shows even better performance than the original cubetree as the query range becomes larger.