Performance fault locating method for distributed database

Abstract

The invention provides a performance fault locating method for distributed database. The method comprises following steps: positioning performance fault nodes which are slow in execution; determining whether the SQL execution plan of the performance fault nodes is changed; if the SQL execution plan is changed, the performance fault positioning is finished and the SQL execution plan of performance fault nodes are optimized, or else, system resource loads, the coordinator performance and user network conditions are successively examined until the performance fault positioning is completed. The method of the invention is simple and efficient; by means of the method, performance faults during user calling database under distributed environment can be quickly and accurately located, greatly reducing the performance positioning time under MPP (distributed database) environment and giving engineers clearer thinking; useless works are avoided and engineers can be more concentrated in solving problems.

Description

technical field [0001] The invention relates to the technical field of distributed databases, in particular to a performance fault location method for distributed databases. Background technique [0002] In recent years, with the development of the Internet and mobile Internet, the amount of data stored by enterprises has increased geometrically, and the traditional way of storing data in databases has become less and less practical, and new data storage architectures have been invented. [0003] MPP storage architect is an extension of the traditional single-node database, which allows data to be stored on more nodes, and does not have much impact on the business logic of the system. It is a big data storage and processing preferred by many enterprises. s solution. [0004] However, the MPP architecture has brought more maintenance problems to the operation and maintenance personnel. The most prominent one is the positioning of performance problems. Since the user's query ...

AI Technical Summary

Problems solved by technology

[0004] However, the MPP architecture has brought more maintenance problems to the operation and maintenance personnel. The most prominent one is the positioning of performance problems. Since the user's query request occurs on multiple database nodes, when a performance problem occurs, it needs to be located on multiple nodes. It is much more difficult to locate the problem on a single database than the performance on a single database. This is a great challenge for the DBA. The more nodes there are in the MPP, the more difficult it is to locate the fault.

Images