Load estimating system and computer load estimating method

• US 7,805,410 B2
• Filed: 03/05/2008
• Issued: 09/28/2010
• Est. Priority Date: 03/08/2007
• Status: Expired due to Fees

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Sasaki, Shigero
• Primary Examiner(s): LEWIS, CHERYL RENEA
• Application Number:
  US12/042,844
• Publication Number:
  US 20080222090A1
• Time in Patent Office: 937 Days
• Field of Search: 707/674, 707/802
• US Class Current: 707/674
• CPC Class Codes: G06F 16/21 Design, administration or maintenance of databases

Abstract

A load estimating system which estimates a load of a database, includes a storage unit, a record number calculating module, and an access number calculating module. The storage unit includes structure information defining a structure of a table in said database, first distribution information indicating probability distribution of condition variables of a query for said table, and second distribution information indicating probability distribution of values in said table. The record number calculating module calculates the average number of selected records which meet conditions defined by-condition variables, based on said structure information, said first distribution information and said second distribution information. The access number calculating module calculates the number of block accesses for said query based on the average number of selected records.

Claims

1. A load estimating system which estimates a load of a database, comprising:

a storage unit configured to include structure information defining a structure of a table in said database, first distribution information indicating probability distribution of condition variables of a query for said table, and second distribution information indicating probability distribution of values in said table;

a record number calculating module configured to calculate the average number of selected records which meet conditions defined by condition variables, based on said structure information, said first distribution information and said second distribution information; and

an access number calculating module configured to calculate the number of block accesses for said query based on the average number of selected records wherein said condition variables define value ranges,

wherein said first distribution information indicates probability distribution of said value ranges,

wherein said record number calculating module calculates an average selection range which is an average value of said value ranges, based on said first distribution information,

wherein said record number calculating module calculates a record range which is an average value of value ranges taken by said values per record, based on the total number of a plurality of records and said second distribution information, and

wherein said record number calculating module calculates the average number of selected records by dividing said average selection range by said record range.

2. The load estimating system according to claim 1, wherein said structure information indicates the total number of said plurality of records in said table and a size of each of said plurality of said records.

3. The load estimating system according to claim 2, wherein said access number calculating module calculates the total number of blocks required to store all of said plurality of records as the number of block accesses based on the total number of said plurality of records, said size of each of said plurality of records and a size of one block, in a full scan.

4. The load estimating system according to claim 1, wherein said storage unit further includes arrangement information indicating a method of arranging said plurality of records into said block and metadata structure information defining a structure of metadata used for an index scan, and

said access number calculating module calculates the number of block accesses based on the average number of selected records, said arrangement information and said metadata structure information.

5. The load estimating system according to claim 4, wherein said arrangement information indicates that said plurality of records is randomly arranged independently of said values,

said access number calculating module calculates the sum of the average number of selected records and the number of accesses to a block where said metadata is stored as the number of block accesses.

6. The load estimating system according to claim 4, wherein said arrangement information indicates that said plurality of records is sequentially arranged in a value order of said values,

when the number of records stored in one block is A1 and the average number of selected records is A2, said access number calculating module calculates a value A3 based on an equation of A3=1+(A2−

1)/A1, and further calculates the sum of said value A3 and the number of accesses to a block where said metadata is stored as the number of block accesses.

7. The load estimating system according to claim 4, wherein said access number calculating module calculates the total number of blocks required to store all of said plurality of records as the number of block accesses based on the total number of said plurality of records, said size of each of said plurality of records and a size of one block, in a full scan.

8. The load estimating system according to claim 7, wherein said arrangement information indicates that said plurality of records is randomly arranged independently of said values,

said access number calculating module calculates the sum of the average number of selected records and the number of accesses to a block where said metadata is stored as the number of block accesses.

9. The load estimating system according to claim 7, wherein said arrangement information indicates that said plurality of records is sequentially arranged in a value order of said values,

when the number of records stored in one block is A1 and the average number of selected records is A2, said access number calculating module calculates a value A3 based on an equation of A3=1+(A2−

1)/A1, and further calculates the sum of said value A3 and the number of accesses to a block where said metadata is stored as the number of block accesses.

10. The load estimating system according to claim 1, further comprising:

a display device configured to display the number of block accesses.

11. The load estimating system according to claim 1, further comprising:

a performance estimating module,

wherein said storage unit further includes load information indicating a computer load required to access one block,

wherein said performance estimating module calculates a load necessary for a process of said query, based on said computer load and the number of block accesses.

12. The load estimating system according to claim 11, further comprising:

a display device configured to display said load necessary for the process of said query.

13. A method of estimating a computer load, which estimates a load of a database, comprising:

reading structure information defining a structure of a table in said database from a storage unit;

reading first distribution information indicating probability distribution of condition variables of a query for said table from said storage unit;

reading second distribution information indicating probability distribution of values in said table from said storage unit;

calculating the average number of selected records which meet conditions defined by condition variables, based on said structure information, said first distribution information and said second distribution information; and

calculating the number of block accesses for said query based on the average number of selected records

wherein said structure information indicates the total number of a plurality of records in said table and a size of each of said plurality of said records,

wherein said condition variables define value ranges,

wherein said first distribution information indicates probability distribution of said vale ranges,

wherein said step of calculating the average number of selected records, includes;

calculating an average selection range which is an average value of said value ranges, based on said first distribution information,

calculating a record range which is an average value of value ranges taken by said values per record, based on the total number of said plurality of records and said second distribution information, and

calculating the average number of selected records by dividing said average selection range by said record range.

14. The method of estimating a computer load according to claim 13, wherein said step of calculating the number of block accesses, includes:

reading out arrangement information indicating a method of arranging said plurality of records into said block and metadata structure information defining a structure of metadata used for an index scan from said storage unit, and

calculating the number of block accesses based on the average number of selected records, said arrangement information and said metadata structure information.

15. A computer program product for a method of estimating a computer load, which estimates a load of a database, embodied on a computer-readable medium and comprising code that, when executed, causes a computer to perform the following:

reading structure information defining a structure of a table in said database from a storage unit;

reading first distribution information indicating probability distribution of condition variables of a query for said table from said storage unit;

reading second distribution information indicating probability distribution of values in said table from said storage unit;

calculating the average number of selected records which meet conditions defined by condition variables, based on said structure information, based on said structure information, said first distribution information and said second distribution information; and

calculating the number of block accesses for said query based on the average number of selected records

wherein said structure information indicates the total number of a plurality of records in said table and a size of each of said plurality of said records,

wherein said condition variables define vale ranges,

wherein said first distribution information indicates probability distribution of said value ranges,

wherein said step of calculating the average number of selected records, includes;

calculating an average selection range which is an average value of said value ranges, based on said first distribution information,

calculating a record range which is an average value of value ranges taken by said values per record, based on the total number of said plurality of records and said second distribution information, and

calculating the average number of selected records by dividing said average selection range by said record range.

16. The computer program product according to claim 15, wherein said step of calculating the number of block accesses, includes:

reading out arrangement information indicating a method of arranging said plurality of records into said block and metadata structure information defining a structure of metadata used for an index scan from said storage unit, and

calculating the number of block accesses based on the average number of selected records, said arrangement information and said metadata structure information.