Systems, clients, servers, methods, and computer-readable media of varying scope are described in which, a database client applies an adaptive method to dynamically determines whether a particular request should execute on the client-side or the server-side of a client-server database system. In determining where a particular request should be executed, the database client analyzes the size of the data sets involved and the data flow generated by the data sets.

A method, apparatus, and software are disclosed for analyzing the data in an OLAP or other type of database. The data is analyzed as if it were indexed by desired dimensions without actually creating the dimensions. Relations known as virtual dimensions are defined in relation to existing base dimensions in the database rather than in relation to the underlying detailed data. Functions operating on these virtual dimensions are translated into functions that operate on the existing base dimensions. As a result, the need to perform certain precalculations is avoided, thereby promoting efficient use of computing resources.

A method, apparatus, and software are disclosed for efficient selection of aggregations, or summaries of data in a database, for answering queries. Benefit/cost ratings are maintained for possible aggregations and are used in determining which aggregations should be selected. Lists of candidate and selected aggregations are also maintained, and aggregations are moved between these lists based on their benefit/cost ratings. These ratings are adjusted as aggregations are moved between the lists.

A multidimensional storage model includes a set of non-sparse entries for each of a plurality of dimensions. The non-sparse entries each identify an associated data value. A set of interdimensional links is provided for each non-sparse entry. The interdimensional links each identify an intersection between non-sparse entries in disparate dimensions and collectively identify all intersections between non-sparse entries in the dimensions.

Techniques for simplifying the aggregation of data stored in a number of objects in a database system. A user-defined type and an object view are associated with a kind of data to be aggregated. The object view specifies the locations of the data in the objects and maps the data to be aggregated into the user-defined type. An application program interface includes a fetch interface that causes the database system to make a query over the object view associated with the kind of data and return the aggregated data in a set of objects of the user-defined type. The returned data is converted to XML and may be further transformed using XSL stylesheets. The application program interface further includes a filter interface that specifies a restriction for the query, a count interface that specifies the number of objects to be returned at once by the application program interface, a transform interface that specifies an XSL style sheet, and a transform parameter interface for providing parameters to the XSL style sheet. The techniques may be used to aggregate metadata for various classes of objects from a data dictionary in a database system. In this application, each class of object has a name and a user-defined type and object view are associated with the class name. The fetch interface specifies a class name and the user-defined type and object view associated with the name are used to aggregate the metadata. One of the transforms performed in this application is the transformation of the metadata for an object into creation DDL for the object.