Method and apparatus for data validation in multidimensional database

• US 20050038799A1
• Filed: 08/11/2004
• Published: 02/17/2005
• Est. Priority Date: 08/11/2003
• Status: Active Grant

• Current Assignee: Descisys Limited
• Original Assignee: Descisys Limited
• Inventors: Jordan, Philip Martin, Stewart, Jerry, Ng, Keith Steffen, Sanders, Michael John
• Granted Patent: US 7,366,725 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/283 Multi-dimensional databases or data warehouses, e.g. MOLAP or ROLAP
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A method of managing cells formed from a tuple of members in a multidimensional database include treating at least one dimension in the database as a plurality of hierarchically structured members divided into at least two levels, thereby defining a validation curve as the location of an intersection of the levels. What level the cells in the database are located is evaluated. Cells contained in a lower level are managed by automatically indicating them as one of valid and invalid at data load time, and cells contained in an upper level are managed by indicating them as one of valid and invalid according to a method involving at least one of less reading from and writing to the database at data load time than if all the cells in the upper level were indicated as one of valid and invalid at data load time. A computerized database management system uses a communication interface to practice the method.

Claims

1. A method of managing cells formed from a tuple of members in a multidimensional database, the method comprising treating at least one dimension in the database as a plurality of hierarchically structured members divided into at least two levels, thereby defining a validation curve as the location of an intersection of said levels, evaluating in what level cells in the database are located and managing cells contained in a lower level by automatically indicating them as one of valid and invalid at data load time and managing cells contained in an upper level by indicating them as one of valid and invalid according to a method involving at least one of less reading from and writing to the database at data load time than if all the cells in the upper level were indicated as one of valid and invalid at data load time.

2. The method of claim 1, wherein the location of at least one validation curve for at least one dimension varies across the dimension.

3. The method of claim 1, wherein the multidimensional database comprises at least two dimensions in the database treated as a plurality of hierarchically structured members having at least two levels and the method comprises locating the validation curve within each dimension separately.

4. The method of claim 1, further comprising allowing for the adjustment of the location of the validation curve of at least one dimension.

5. The method of claim 4, further comprising automatically adjusting the operation of the multidimensional database by adjusting the location of the validation curve according to historical data.

6. The method of claim 1, wherein each cell comprises at least two members and wherein the method further comprises allowing for the adjustment of the method by which cells are determined to be above or below the validation curve.

7. The method of claim 6, further comprising automatically adjusting the method by which cells are determined to be above or below the validation curve according to historical data.

8. The method of claim 7, wherein the historical data indicates the location of cells in the database that have been queried, used as part of a larger calculation and/or updated and the frequency that cells in the database have been queried, used as part of a larger calculation and/or updated.

9. The method of claim 1, further comprising treating the at least one dimension in the database as a plurality of hierarchically structured members divided into two levels, consisting of one upper level and one lower level.

10. The method of claim 1, wherein cells in said upper level are managed according to at least one of the options:

a) the cells are permanently indicated as invalid;

b) the cells are stored with a timestamp indicating the time that the cell was last updated, wherein a calculated cell is evaluated as being one of valid and invalid by comparing the timestamp of the calculated cell with the timestamp of its source cells located above the validation curve and by determining the validity of its source cells located below the validation curve; and

c) only cells that may be queried are stored in the multidimensional database and the cells are indicated as one of valid and invalid at data load time.

11. The method of claim 10, wherein all cells in said upper level for all dimensions are managed according to one of the options a)-c).

12. The method of claim 1, wherein no cells above the validation line are stored in the multidimensional database.

13. The method of claim 10, wherein option b) is used and if the calculated cell is evaluated to be invalid the method comprises recalculating the calculated cell by evaluating the validity of its source cells, working down the hierarchical structure until a base of valid cells is reached and then working up the hierarchical structure recalculating cells until said calculated cell is recalculated.

14. The method of claim 1, wherein the multidimensional database comprises an index defining its cells, the index including a pointer to the relevant data forming the cells, wherein the validity of a cell is indicated by associating validity information with the index.

15. The method of claim 14, wherein the cells are formed from at least two members from different dimensions.

16. A computerised database management system for a multidimensional database comprising cells formed from a tuple of members, the computerised database management system comprising a communication interface for reading from and writing to a database, computer processing means for controlling the operations of the database management system and computer memory containing an instruction set readable by said computer processing means, wherein when the instruction set is executed by the computer processing means the computerised database management system uses said communication interface to:

a) maintain at least one dimension in the database as a plurality of hierarchically structured members divided into at least two levels, thereby defining a validation curve as the location of an intersection of said levels; and

b) evaluate in what level cells in the database are located and managing cells contained in a lower level by automatically indicating them as one of valid and invalid at data load time and managing cells contained in an upper level by indicating them as one of valid and invalid according to a method involving at least one of less reading from and writing to the database at data load time than if all the cells in the upper level were indicated as one of valid and invalid at data load time.

17. The computerised database management system of claim 16 including a pluggable component operable to set the location of the validation curve.

18. The computerised database management system of claim 17, wherein said pluggable component is operable to analyse historical data on the use of the multidimensional database and automatically adjust the location of the validation curve dependent on the historical data.

19. The computerised database management system of claim 16 including a pluggable component operable to set a method by which cells are determined to be above or below the validation curve, by setting the number of members that need to be above or below the validation curve in order for the cell to be above or below the validation curve.

20. The computerised database management system of claim 19, wherein said pluggable component is operable to analyse historical data on the use of the multidimensional database and automatically adjust the location of the validation curve dependent on the historical data.

21. A method of managing cells formed from a tuple of members in a multidimensional database, the method comprising maintaining an index to the cells, the index defining a hierarchical structure containing pointers to the members that make up each cell, allocating each cell to one of at least two levels in the hierarchical structure and associating with the index a cell validity indicator for each cell, wherein the cell validity indicator is updated upon the occurrence an event, wherein the class of event that causes the cell validity indicator to be updated is dictated by the level in which the cell is allocated.

22. The method of claim 21, comprising automatically updating the cell validity indicator for cells allocated to the lowermost level when new data is loaded into the database.

23. The method of claim 21, wherein cells in an upper level of said at least two levels are permanently indicated as invalid.

24. The method of claim 21, wherein the cells in an upper level of said at least two levels are stored with a timestamp indicating the time that the cell was last updated and wherein a calculated cell is evaluated as being one of valid and invalid by comparing the timestamp of the calculated cell with the timestamp of its source cells in the upper level and by determining the validity of its source cells that are located below the validation curve.

25. The method of claim 21, wherein only cells that may be queried are stored in the multidimensional database and the cells are indicated as one of valid and invalid at data load time.