Data base optimizer using most frequency values statistics

• US 4,956,774 A
• Filed: 09/02/1988
• Issued: 09/11/1990
• Est. Priority Date: 09/02/1988
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Shibamiya, Akira, Zimowski, Melvin R.
• Primary Examiner(s): NOT, DEFINED
• Application Number:
  US07/239,712
• Time in Patent Office: 739 Days
• Field of Search: 364/200, 364/300, 364/900
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24549 Run-time optimisation
• CPC Class Codes: G06F 16/9017 using directory or table look-up use of a directory or look-up table in file systems G06F16/13
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching

Abstract

A method for more accurately estimating the time required to process a data base query using a selected index. A selected number of the most frequently occurring index key values (38) are collected during an index sequential scan. These most frequency occurring values are stored as percentage frequencies of occurrence in the data base system'"'"'s catalog (42). Estimated access and processing times (NPAR, NPAS, NCPU) for a given query are calculated based on the stored frequencies where possible. Where the query'"'"'s search criteria specify values other than the stored most frequently occurring values, those values are assumed to be uniformly distributed.

Claims

1. In a method for accessing data of a relational data base management system having at least one index, the improvement characterized by the steps performed by a computer of:

(a) selecting a number of most frequently occurring values of at least part of a key of the index, the number being greater than zero and less than a total number of such values;

(b) collecting frequency of occurrence statistics for the selected most frequently occurring values of the index;

(c) estimating a time required for using the index as the access path, based at least in part on the collected frequency of occurrence statistics;

(d) selecting an access path based at least in part on the estimated time; and

(e) accessing the data using the selected access path.

2. The method of claim 1, wherein frequencies of occurrence of the values other than the collected most frequently occurring values are assumed to be uniformly distributed.

3. The method of claim 1, wherein the statistics comprise the most frequently occurring values, and their respective percentage frequencies of occurrence.

4. The method of claim 1, wherein the number of most frequently occurring values is selected so that a sum of frequencies of occurrence of the selected most frequently occurring values is greater than a selected threshold.

5. The method of claim 1, wherein the number of most frequently occurring values is selected to include all values occurring with more than a selected threshold frequency as most frequently occurring values.

6. The method of claim 1, wherein the collected statistics are stored within the data base management system.

7. The method of claim 6, wherein the collected statistics are stored in a system catalog.

8. The method of claim 1, wherein times required for random and sequential page accesses and for processing time are estimated separately, based at least in part on the index'"'"'s statistics.

9. A method for accessing data of a relational data base management system having at least one index, comprising the steps performed by a computer of:

(a) selecting a number of most frequently occurring values of at least part of a key of the index, the number being greater than zero and less than a total number of values of said at least part of the key;

(b) collecting frequency of occurrence statistics for the selected most frequently occurring values of the index;

(c) storing the statistics in the system;

(d) estimating a time required for using the index as the access path, based at least in part on the stored frequency of occurrence statistics;

(e) selecting an access path based at least in part on the estimated time; and

(f) accessing the data using the selected access path.