Implementing data cubes efficiently-英文文献.pdf

Implementing Data Cubes Efficiently* Venky Harinarayan Anand Rajaraman Jeffrey D. U1lman Stanford University Stanford University Stanford IJniversity venky@cs. st anford. edu anand@cs. st anford. edu ullman@cs .st anford. edu Abstract databases called data warehouses on which users can carry out their analysis. Decision support applications involve complex queries on very large databases. Since response times should While operational databases maintain state infor- be small, query optimization is critical. Users typically mation, data warehouses typically maintain histori- view the data as multidimensional data cubes. Each cell cal information. As a result, data warehouses tend of the data cube is a view consisting of an aggregation to be very large and to grow over time. LTsers of of interest, like total sales. The values of many of these DSS are typically interested in identifying trends cells are dependent on the values of other cells in the rather than looking at individual recor