Database query optimization method based on reinforcement learning and graph attention network

Abstract

The invention relates to the technical field of databases, provides a database query optimization method based on reinforcement learning and a graph attention network, and aims at solving the technical problems that when the connection relationship of existing query statements is very complex, the query execution plan space is very huge, and a large amount of time is consumed for searching for thewhole query execution space. According to the main scheme, the method comprises the steps of: randomly generating a query statement in a database and executing the query statement, splitting an execution plan tree corresponding to the query statement from a root node, and recording a connection relationship of each node; initializing a Qnetwork parameter w in the DQN model, enabling a Qnetwork inthe DQN model to employ a GAT graph attention network, enabling the coding feature matrix and the graph description set Edge to serve as network input, and training the DQN model; and for a query statement, initializing graph description and coding of the query statement, and generating a connection relationship by using the DQN model obtained by training in the step 2 until all tables are connected to generate a complete query plan.

Description

technical field [0001] The invention relates to a database query optimization method based on reinforcement learning and graph attention network. For large-scale multi-connection queries, an optimal database query execution plan can be obtained in a short period of time, thereby reducing the execution time of queries in the database. Background technique [0002] For a query statement, the database cannot execute it directly. The database needs to first parse the query statement, then the optimizer generates the corresponding query execution plan, and finally hand it over to the execution engine to execute the plan. In the present invention, an effective solution is proposed on how to generate a better query plan for multi-connection queries in a shorter period of time. [0003] The technical solutions in the two existing technologies are closest to the proposal of this application: [0004] 1. Chinese invention patent, patent name: a database multi-connection query optim...

AI Technical Summary

Problems solved by technology

[0029]The purpose of this invention is to solve the problem that when the connection relationship of existing query statements is very complicated, the query execution plan space will be very large, and it will take a lot of time to search the entire query execution space technical problem

Images