Method of changing system configuration in shared-nothing database management system

• US 20050154731A1
• Filed: 04/30/2004
• Published: 07/14/2005
• Est. Priority Date: 01/09/2004
• Status: Active Grant

• Current Assignee: Avago Technologies General IP PTE Limited (Broadcom, Inc.)
• Original Assignee: Avago Technologies General IP PTE Limited (Broadcom, Inc.)
• Inventors: Ito, Daisuke, Fujiwara, Shinji, Ushijima, Kazutomo, Maciel, Frederico Buchholz
• Granted Patent: US 7,370,029 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/284 Relational databases
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface

Abstract

A shared-nothing database management system employs a method of adding and removing a database server without significantly affecting the processing performance for accessing a table from a user or an application. A scheduler module creates a schedule which separately handles the addition of a CPU resource and the addition of a storage I/O resource, and a server is added in accordance with the created schedule. A common disk is used to remove a disk at an arbitrary timing. Further, a data region is previously subdivided on the common disk to eliminate a movement of data associated with the addition of a server.

Claims

1. A system configuration changing method in a shared-nothing database management system which includes a plurality of CPU resources, a storage, and a network for connecting said CPU resources and said storage, wherein at least some of said plurality of CPU resources run a database search program to each function as a database server for processing data to be processed relatively prime to one another, said method comprising the steps of:

adding a server exclusively for sorting and merging data without accessing a table in a database to said shared-nothing database management system;

determining whether or not a storage I/O resource is added;

changing said server exclusively for sorting and merging data to an ordinary server for accessing a table in a database in response a determination that the storage I/O resource is added; and

relocating data when said database server is lightly loaded.

2. A system configuration changing method according to claim 1, wherein said step of adding a server exclusively for sorting and merging data includes activating a program for sorting and merging data on a pooled server which is connected to said network and is not functioning as a CPU resource for said shared-nothing database management system.

3. A system configuration changing method in a shared-nothing database management system which includes a plurality of CPU resources, a storage, and a network for connecting said CPU resources and said storage, wherein at least some of said plurality of CPU resources run a database search program to each function as a database server for processing data to be processed relatively prime to one another, said method comprising the steps of:

monitoring the state of the CPU resources functioning as said database servers to determine removal of a database server from said shared-nothing database management system;

removing the database server determined to be removed without delay, and delegating a data region previously allocated to the database server for storing data to be processed to another database server;

relocating data on the delegated storage region to another storage region; and

deleting the delegated storage region from which actual data has been moved.

4. A system configuration changing method in a shared-nothing database management system which includes a plurality of CPU resources, a storage, and a network for connecting said CPU resources and said storage, wherein at least some of said plurality of CPU resources run a database search program to each function as a database server for processing data to be processed relatively prime to one another, said method comprising the steps of:

setting a correspondence table indicative of a correspondence of respective data regions subdivided on said storage to database servers responsible for processing to determine the allocation of a plurality of data regions for storing data to be processed to the respective database servers;

determining whether a database server should be added or removed; and

rewriting part of said correspondence table to determine allocation of new data to be processed by a plurality of database servers after addition or removal.

5. A shared-nothing database management system which includes a plurality of CPU resources, a storage, and a network for connecting said CPU resources and said storage, wherein at least some of said plurality of CPU resources run a database search program to each function as a database server for processing data to be processed relatively prime to one another, wherein:

said storage comprises a storage virtualization layer for making a plurality of unit volumes available as each virtual volume for each of said plurality of database servers, an interface for notifying said database server of the capacity of said unit volume, and an interface for changing the configuration of said virtual volumes in response to a request from said database server; and

said database search program comprises an interface mechanism for querying said storage the capacity of said unit volume, a mechanism for placing data groups in respective unit volumes without overlapping, and an interface mechanism for requesting said storage to change the configuration of the virtual volumes.

6. A shared-nothing database management system according to claim 5, wherein said storage virtualization layer is installed on a management middleware of said storage;

7. A non-share database management system comprising a plurality of database server and an apparatus for replacing a movement of data between database servers with a change in virtual volumes of a file system, said apparatus comprising:

1. a storage shared on a network;

2. a file system for making a plurality of unit volume available as virtual volumes;

3. a database search program having a mechanism for assigning a data movement to a file system,

wherein said storage has;

a. an interface for notifying the file system of the capacity of the unit volume,

said file system has;

i. a storage virtualization layer for making a plurality of unit volumes available as virtual volumes;

ii. an interface for notifying the database servers of the capacity of the unit volume; and

iii. an interface for changing the configuration of the virtual volumes in response to a request from a database server, and

said database search program runs functionally as;

A. a data group which defines a minimum unit of movement in the event of a data movement;

B. an interface for querying the file system each unit volume;

C. a mechanism for placing data groups in the unit volumes without overlapping; and

D. an interface for requesting the file system to change the configuration of the virtual volumes.

8. A shared-nothing database management system comprising a plurality of database servers, and an apparatus for replacing a data movement between database servers with a passage of unit volumes, said apparatus comprising:

1. a storage shared on a network; and

2. a database search program having a mechanism for virtually moving data,

wherein said storage comprises;

a. an interface for notifying the database servers of the capacity of the unit volume, and

said database search program has;

i. a data group which defines a minimum unit of movement in the event of a data movement;

ii. an interface for querying the file system each unit volume;

iii. a mechanism for placing data groups in the unit volumes without overlapping; and

iv. a mechanism for changing the configuration of the unit volumes.