Time index access structure for temporal databases having concurrent multiple versions

• US 5,440,730 A
• Filed: 08/09/1990
• Issued: 08/08/1995
• Est. Priority Date: 08/09/1990
• Status: Expired due to Term

• Current Assignee: TTI Inventions C LLC (Intellectual Ventures LLC)
• Original Assignee: Bell Communications Research, Inc. (Telefonaktiebolaget LM Ericsson)
• Inventors: Wuu, Tzyh-Jain G., Elmasri, Ramez A.
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): Lintz, Paul R.
• Application Number:
  US07/564,881
• Time in Patent Office: 1,825 Days
• Field of Search: 395/600, 395/650, 395/725, 395/425, 375/107, 364/419.19
• US Class Current: 1/1
• CPC Class Codes: G06F 16/00 Information retrieval; Database structures therefor; File system structures therefor
• CPC Class Codes: Y10S 707/99954 Version management

Abstract

A time index for temporal databases is provided which enables the retrieval of database object versions that are valid during specified time periods. Unlike prior access and retrieval structures, the present index is based on objects whose search values are time intervals rather than time points. A series of ordered indexing points is defined by the start and end of object version intervals and these points are used to build an indexing structure, which may take the form of a B⁺ -tree. Each leaf node entry of the B⁺ -tree represents an indexing point and has an associated bucket of pointers which identify all object versions that are valid at that time. Storage space is reduced by including only incremental change indicators in the buckets of non-leading leaf entries and calculating needed pointers from such indicators. The time index may be employed in multi-level structures with attribute indexes to greatly improve the efficiency of temporal search operations, such as aggregate functions and temporal selection, as well WHEN and JOIN operators.

Claims

1. A computer-based temporal database management system including a time index which comprises:

a) an ordered series of indexing time points defining time intervals during which at least one of a plurality of concurrent object version in said database is valid; and

b) associated with each indexing point, pointer means identifying all database object versions that are valid at the time represented by said each indexing point.

2. A system according to claim 1 wherein said each indexing time point defines a change of state of said database with respect to an object version.

3. A system according to claim 1 wherein said indexing time points are situated on a scale of regularly occurring time points and represent those scalar time points:

a) at which a database object version begins; and

b) next following the scalar time point at which a database object version terminates.

4. A system according to claim 1 wherein said indexing time points populate the leaf nodes of a B⁺ -tree index structure.

5. A system according to claim 4 wherein each said indexing time point has an associated pointer to a bucket of pointers identifying said valid database object versions.

6. A system according to claim 5 wherein each said valid database object version is identified by a bucket pointer.

7. A system according to claim 5 wherein the bucket pointers of selected indexing points identify only beginning and terminating database object versions.

8. A method for structuring a time index in a computer-based temporal database management system which comprises:

a) assembling in computer memory an ordered series of indexing time points defining time intervals during which at least one of a plurality of concurrent object version in said database is valid; and

b) associating with each indexing point pointer means identifying all database object versions that are valid at the time represented by said each indexing point.

9. A method according ,to claim 8 which further comprises selecting as said indexing time points those points on a scale of regularly occurring time points that define a change of state of said database with respect to an object version.

10. A method according to claim 9 which comprises selecting time points:

a) at which a database object version begins; and

b) next following the scalar time point at which a database object version terminates.

11. A method according to claim 8 wherein said assembling comprises inserting said indexing time points as the population of the leaf nodes of a B⁺ -tree index structure.

12. A method according to claim 11 which further comprises associating with each said indexing time point a pointer to a bucket of pointers identifying said valid database object versions.

13. A method according to claim 12 which further comprises associating with said time index B⁺ -tree, by means of a leaf node pointer therefrom to said time index B⁺ -tree, a higher level B⁺ -tree index of database attributes related to the time interval object versions of said database.

14. A method according to claim 12 which further comprises, upon inserting in said database a new object version whose time interval extends from a first time point to a second time point:

a) if there exists no indexing point at said first time point, creating an indexing point at said first time point and creating a pointer bucket for said indexing point;

b) if there exists no indexing point at the next scalar time point following said second time point, creating an indexing point at said next time point and creating a pointer bucket for said indexing point; and

c) inserting a pointer to said new object version in the bucket for the indexing point at said first time point and in the buckets for all subsequent indexing points prior to said second time point.