Method and apparatus for predicting selectivity of database query join conditions using hypothetical query predicates having skewed value constants

• US 7,987,200 B2
• Filed: 10/31/2007
• Issued: 07/26/2011
• Est. Priority Date: 11/18/2004
• Status: Expired due to Fees

• Current Assignee: SAP SE
• Original Assignee: International Business Machines Corporation
• Inventors: Muras, Brian Robert, Kethireddy, Shantan, Faunce, Michael S.
• Primary Examiner(s): Mofiz; Apu M
• Assistant Examiner(s): NGUYEN, THU N
• Application Number:
  US11/931,255
• Publication Number:
  US 20080052286A1
• Time in Patent Office: 1,364 Days
• Field of Search: None
• US Class Current: 707/791
• CPC Class Codes: G06F 16/24545 Selectivity estimation or determination
• CPC Class Codes: G06F 16/24549 Run-time optimisation
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation
• CPC Class Codes: Y10S 707/99942 Manipulating data structure, e.g. compression, compaction, compilation
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A database management system predicts a selectivity for database query conditions requiring a join of records from different tables. The system identifies at least one skewed value in a field specified in the join condition, and constructs, for each skewed value, a set of hypothetical query predicates in which the field specified in the join condition is replaced with a constant equal to the skewed value. The system then predicts the selectivity for the hypothetical predicates, using any appropriate prediction technique. The selectivities of the hypothetical predicates are used to predict a selectivity for the original query.

Claims

1. A method for predicting the selectivity of a set of at least one logical query condition for querying a database, said at least one logical query condition comprising at least one join condition for joining multiple tables of said database, comprising the steps of:

automatically identifying at least one skewed value for a first field of said database specified in said at least one join condition;

for each skewed value identified by said step of automatically identifying at least one skewed value, automatically constructing a corresponding set of one or more hypothetical query predicates, each hypothetical query predicate of said set of hypothetical query predicates corresponding to a respective query condition of said set of at least one logical query condition, wherein each respective said hypothetical query predicate is constructed by replacing each occurrence of said first field in the corresponding query condition with a constant equal to the skewed value corresponding to the set of hypothetical query predicates containing the hypothetical query predicate;

automatically predicting a respective selectivity for each said set of hypothetical query predicates; and

automatically predicting a composite selectivity for said set of at least one logical query condition using the respective selectivity for each said set of hypothetical query predicates;

automatically identifying at least one skewed value for a first field comprises automatically accessing a frequent value list containing sampled values from said first field, and automatically identifying any sampled values exceeding at least one pre-determined threshold;

automatically identifying any sampled values exceeding at least one pre-determined threshold comprises automatically identifying any sampled values exceeding at least two pre-determined thresholds, a first of said pre-determined thresholds representing a number of occurrences of said sampled value, and a second of said pre-determined thresholds representing a proportion of records containing said sampled value among all records in a database table containing said first field.

2. The method of claim 1, wherein said method is performed by a database management system for purposes of determining an optimum query execution strategy for executing a logical query of said database.

3. The method of claim 1, wherein said step of automatically predicting a respective selectivity for each said set of hypothetical query predicates comprises automatically accessing at least one respective index for each of at least one hypothetical query predicate, and automatically predicting a number of records satisfying each said at least one hypothetical query predicate using said respective index.

4. A computer storage media for predicting the selectivity of a set of at least one logical query condition for querying a database, said at least one logical query condition comprising at least one join condition for joining multiple tables of said database, said computer storage media comprising:

a plurality of computer-executable instructions stored on said computer storage media, wherein said instructions, when executed by at least one computer system, cause the at least one computer system to perform the steps of;

(a) identifying at least one skewed value for a first field of said database specified in said at least one join condition;

(b) for each skewed value identified by said step of identifying at least one skewed value, constructing a corresponding set of one or more hypothetical query predicates, each hypothetical query predicate of said set of hypothetical query predicates corresponding to a respective query condition of said set of at least one logical query condition, wherein each respective said hypothetical query predicate is constructed by replacing each occurrence of said first field in the corresponding query condition with a constant equal to the skewed value corresponding to the set of hypothetical query predicates containing the hypothetical query predicate;

(c) predicting a respective selectivity for each said set of hypothetical query predicates; and

(d) predicting a composite selectivity for said set of at least one logical query condition using the respective selectivity for each said set of hypothetical query predicates;

identifying at least one skewed value for a first field comprises accessing a frequent value list containing sampled values from said first field, and identifying any sampled values exceeding at least one pre-determined threshold;

identifying any sampled values exceeding at least one pre-determined threshold comprises identifying any sampled values exceeding at least two pre-determined thresholds, a first of said pre-determined thresholds representing a number of occurrences of said sampled value, and a second of said pre-determined thresholds representing a proportion of records containing said sampled value among all records in a database table containing said first field.

5. The computer storage media of claim 4, wherein said computer storage media embodies a database management system which executes logical queries against said database, said database management system predicting selectivity of a set of at least one logical query condition for purposes of determining an optimum query execution strategy for executing a logical query of said database.

6. The computer storage media of claim 4, wherein said step of predicting a composite selectivity for said set of at least one logical query condition comprises predicting a residual selectivity for said set of at least one logical query condition, said residual selectivity being predicted as a selectivity of a hypothetical database which does not contain said skewed values, and combining said residual selectivity with said respective selectivity for each said set of hypothetical query predicates to produce said composite selectivity.

7. The computer storage media of claim 6, wherein said residual selectivity is calculated as a quotient, the numerator of said quotient being the minimum of a plurality of adjusted cardinalities, each adjusted cardinality being a representation of cardinality of a respective database table joined by said at least one join condition less a number of skewed values, if any, identified in the respective database table by said identifying step, and the denominator of said quotient being the product of said plurality of adjusted cardinalities.

8. A computer system, comprising:

at least one processor;

an addressable memory coupled to said at least one processor;

a database containing data storable in said addressable memory, said database comprising a plurality of tables a database management system embodied as a plurality of instructions storable in said addressable memory and executable on said at least one processor, said database management system supporting the execution of logical queries against data in said database, at least some of said logical queries including at least one respective join condition for joining multiple tables of said database;

wherein said database management system estimates selectivity of a set of at least one logical query condition including at least one join condition by

(a) identifying any skewed values in a field specified in said at least one join condition,

(b) constructing a respective set of one or more hypothetical query predicates corresponding to each identified skewed value, each hypothetical query predicate of a set of hypothetical query predicates corresponding to a respective query condition of said set of at least one logical query condition, wherein each respective said hypothetical query predicate is constructed by replacing each occurrence of the field containing the skewed value in the corresponding query condition with a constant equal to the skewed value corresponding to the set of hypothetical query predicates containing the hypothetical query predicate,

(c) predicting a respective selectivity for each said set of hypothetical query predicates, and

(d) predicting a composite selectivity for said set of at least one logical query condition using the respective selectivity for each said set of hypothetical query predicates;

identifying at least one skewed value for a first field comprises accessing a frequent value list containing sampled values from said first field, and identifying any sampled values exceeding at least one pre-determined threshold;

identifying any sampled values exceeding at least one pre-determined threshold comprises identifying any sampled values exceeding at least two pre-determined thresholds, a first of said pre-determined thresholds representing a number of occurrences of said sampled value, and a second of said pre-determined thresholds representing a proportion of records containing said sampled value among all records in a database table containing said first field.

9. The computer system of claim 8, wherein said database further comprises a plurality of indexes, each index corresponding to one or more respective fields within a respective table of said plurality of database tables, and wherein said database management system estimates a respective selectivity for each said set of hypothetical query predicates by accessing at least one respective index for each of at least one hypothetical query predicate, and predicting a number of records satisfying each said at least one hypothetical query predicate using said respective index.