DATA PARTITIONING METHOD OF DISTRIBUTED PARALLEL DATABASE SYSTEM

• US 20120109888A1
• Filed: 12/14/2011
• Published: 05/03/2012
• Est. Priority Date: 07/28/2010
• Status: Abandoned Application

• Current Assignee: Borqs Wireless Limited
• Original Assignee: Beijing Borqs Software Technology Co., Ltd. (Borqs Technologies, Inc.)
• Inventors: ZHANG, Weiping, Zhang, Songbo, Liu, Weihuai
• Application Number:
  US13/325,810
• Publication Number:
  US 20120109888A1
• Time in Patent Office: Days
• US Class Current: 707/610
• CPC Class Codes: G06F 16/278 Data partitioning, e.g. horizontal or vertical partitioning

Abstract

A data partitioning method for a distributed parallel database system, comprising creating fact tables and dimension tables according to a constructed distributed parallel database system, inserting records of the dimension tables and the fact tables into nodes according to partitioning rules, replicating the records of dimension tables into the nodes that include fact tables, performing data deletion, and performing data update.

Claims

1. A data partitioning method of a distributed parallel database system, the data partitioning method comprising:

creating fact tables and dimension tables according to a constructed distributed parallel database system and distribution rules;

inserting records of the fact tables and records of the dimension tables into nodes;

replicating the records of the dimension tables to nodes of the fact tables;

performing data deletion; and

performing a data update.

2. The data partitioning method of a distributed parallel database system according to claim 1, wherein the fact table comprises a primary key, a foreign key, and the records of the fact table.

3. The data partitioning method of a distributed parallel database system according to claim 1, wherein the dimension table comprises a primary key and the records of the dimension table.

4. The data partitioning method of a distributed parallel database system according to claim 1, wherein the fact tables and dimension tables are related with a primary key and a foreign key, and wherein a value of the foreign key of the fact table is equal to a value of the primary key of a related dimension table.

5. The data partitioning method of a distributed parallel database system according to claim 1, wherein said inserting records of fact tables and records of dimension tables into nodes comprises inserting the records of the fact tables and the records of the dimension tables into different nodes.

6. The data partitioning method of a distributed parallel database system according to claim 1, wherein said replicating the records of the dimension tables to nodes of the fact tables comprises:

determining related dimension tables according to foreign keys in the fact tables; and

replicating records of the related dimension tables to the node that contains the fact table.

7. The data partitioning method of a distributed parallel database system according to claim 1, wherein said performing data deletion comprises:

deleting the records of the fact tables;

deleting the records of the dimension tables related with the fact tables in the node; and

keeping the records of the dimension tables in a primary node.

8. The data partitioning method of a distributed parallel database system according to claim 1, wherein said performing a data update comprises:

updating records of each dimension table in a certain node;

searching for the fact tables related with the dimension tables; and

updating the related dimension tables in the nodes that contain the fact tables.

9. The data partitioning method of a distributed parallel database system according to claim 1, wherein said performing data update comprises creating a bloom filter table for each dimension table and each node to record a distribution of the records of each dimension table in each node, to find a node that stores a specified record, and to update each dimension table in the node.

10. The data partitioning method of a distributed parallel database system according to claim 9, wherein the bloom filter table is stored in a front-end server or in each node.

11. The data partitioning method of a distributed parallel database system according to claim 1, wherein said creating fact tables, said replicating the records of the dimension tables, said performing data deletion, and said performing a data update are performed by a general purpose processor.