Index adding program of relational database, index adding apparatus, and index adding method

• US 20060004840A1
• Filed: 11/24/2004
• Published: 01/05/2006
• Est. Priority Date: 07/02/2004
• Status: Abandoned Application

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Senda, Toshiaki
• Application Number:
  US10/996,762
• Publication Number:
  US 20060004840A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/2272 Management thereof

Abstract

An index adding program, index adding apparatus, and index adding method of a relational database, that enables dynamic and efficient addition of a new index without stopping service to an application, by contriving a method of reflecting an update log to the index. The index of the relational database is initially generated and an updating log for the relational database is accumulated during the initial index generation. Thereafter, data corresponding to an update request to the relational database is present in the index; the update request is reflected to the index. If the request is not present, update request is stored in the work area. Further, reflecting of an accumulated update log to the initially generated index is sequentially executed, the update log is disregarded, if the data corresponding to the update log is stored in the work area, or alternatively, the update log is reflected to the index, if data corresponding to the update log is not stored in the work area.

Claims

1. An index adding program of a relational database, for executing on a computer comprising the steps of:

initially generating an index of the relational database;

accumulating an update log for said relational database during said initial index generation;

if after the initial index generation, data corresponding to an update request to said relational database is present in the index, reflecting said update request to the index, or alternatively, if data corresponding to said update request is not present in the index, storing said update request in a first work area; and

when sequentially reflecting an accumulated update log to the initially generated index, disregarding the update log if the data corresponding to said update log is stored in a first work area, or alternatively, reflecting said update log to the index if data corresponding to said update log is not stored in the first work area.

2. An index adding program of a relational database according to claim 1, wherein said step of initially generating an index of said relational database, initially generates an index from back up data appropriately acquired in preparation against destruction of said relational database.

3. An index adding program of a relational database according to claim 2, wherein said back up data is configured from;

a dump file in which said relational database has been saved at certain point in time, and an archive log file in which is recorded update contents of said relational database.

4. An index adding program of a relational database according to claim 1, wherein said step of initially generating an index of said relational database comprises the steps of:

extracting key records from said relational database and storing them in a second work area;

merging key records extracted from an update request to said relational database with a corresponding section in said second work area while sorting key records stored in said second work area in accordance with predetermined rules; and

generating an index from the sorted and merged key records.

5. An index adding program of a relational database according to claim 4 comprising a step for, after generation of the index from said sorted and merged key records, until completion of reflection of the accumulated update log to the index, if data corresponding to the update request to said relational database is present in the index, updating said index in accordance with said update request, or alternatively, if data corresponding to said update request is not present in the index, storing the update request in the first work area.

6. An index adding program of a relational database according to claim 4, wherein said step of generating an index from said sorted and merged key records generates an index while disregarding a combination of insertions and deletions to the same index.

7. An index adding apparatus of a relational database comprising:

an initial generation means which initially generates an index of the relational database;

a log accumulating means which accumulates an update log for said relational database during initial index generation by said initial generation device;

an update request processing means which if after the initial index generation by said initial generation device, data corresponding to an update request to said relational database is present in the index, reflects said update request to the index, or alternatively, if data corresponding to said update request is not present in the index, stores said update request in a work area; and

a log reflection means which when sequentially reflecting an accumulated update log to the initially generated index, disregards the update log if the data corresponding to said update log is stored in the work area, or alternatively, reflects said update log to the index if data corresponding to said update log is not stored in the work area.

8. An index adding method of a relational database which executes on a computer comprising the steps of:

initially generating an index of the relational database;

accumulating an update log for said relational database during said initial index generation;

if after the initial index generation, data corresponding to an update request to said relational database is present in the index, reflecting said update request to the index, or alternatively, if data corresponding to said update request is not present in the index, storing said update request in a work area; and

when sequentially reflecting an accumulated update log to the initially generated index, disregarding the update log if the data corresponding to said update log is stored in the work area, or alternatively, reflecting said update log to the index if data corresponding to said update log is not stored in the work area.