Distributed database based on hypergraph division and cluster partitioning method of distributed database

Abstract

The invention discloses a distributed database based on hypergraph division and a cluster partitioning method of the distributed database and relates to a data placement technology of a database in a big data environment. The distributed database comprises a data counting module, a historical data analysis module and a data migration module, wherein an transaction access set is stored in original text and log file data, and an algorithm module set stores mathematical modeling and algorithm solving of a platform integration algorithm module; a data analysis module comprises two processes, namely, a log reading and analysis process and a data modeling and algorithm solving process; the log reading and analysis process of the data analysis module is used for managing user log uploading, preprocessing and format conversion, and a transaction log after analysis is generated; a scheduling module is used for orderly scheduling modeling, algorithm and data migration, analyzing computed results and performing partitioning again. The method combines market demands, log analysis and the algorithm module, and data placement and partitioning of a platform can be efficiently adjusted; the expandability and the maintainability of the database are improved.

Description

technical field [0001] The invention relates to the field of databases under a big data environment, in particular to a distributed database based on hypergraph partitioning and a cluster partitioning method thereof. Background technique [0002] The data centers of modern enterprises are becoming increasingly large, and distributed databases are widely used in enterprise applications to provide effective management and support for business activities of enterprises. In addition, online transaction processing is a new requirement for distributed databases in business environments. Simply put, Online Transaction Processing (OLTP) is the automated processing of repetitive, high-volume businesses. With the rapid growth of application scenarios for online transaction processing, a distributed database management system designed for online transaction processing has emerged. By analyzing the historical log files of the system, counting the distribution of transactions in the sy...

AI Technical Summary

Problems solved by technology

[0004] Due to the complexity of data association and the huge scale of cluster partitions, transactions are short, small, and repetitive in query and modification. According to the data placement strategy, a large part of transactions will access data on multiple physical machines, and many The network communication overhead of partition transactions is huge, which will greatly limit the popularity of multi-partition transaction network communication

Images