RL_QOptimizer: A Reinforcement Learning Based Query Optimizer

Abstract

With the current availability of massive datasets and scalability requirements, different systems are required to provide their users with the best performance possible in terms of speed. On the physical level, performance can be translated into queries' execution time in database management systems. Queries have to execute efficiently (i.e. in minimum time) to meet users' needs, which puts an excessive burden on the database management system (DBMS). In this paper, we mainly focus on enhancing the query optimizer, which is one of the main components in DBMS that is responsible for choosing the optimal query execution plan and consequently determines the query execution time. Inspired by recent research in reinforcement learning in different domains, this paper proposes A Deep Reinforcement Learning Based Query Optimizer (RL_QOptimizer), a new approach to find the best policy for join order in the query plan which depends solely on the reward system of reinforcement learning. The experimental results show that a notable advantage of the proposed approach against the existing query optimization model of PostgreSQL DBMS.INDEX TERMS Join ordering problem, query execution plan and query optimization.

Structured query language query join optimization by using rademacher averages and mapreduce algorithms

“…The function 𝑤 is convex, continuous in 𝑅 + , and has first and second derivatives 𝑟 everywhere in its domain, so it is possible to minimize it effectively by using standard convex optimization methods. The euclidean norm is described by utilizing (9). Rademacher algorithm and statistical learning can be assumed that a learning algorithm chooses its hypothesis from certain fixed hypothesis class 𝐻, generalization of error analysis gives theoretical results bounding generalization of error of hypothesis ℎ ∈ 𝐻 which will be based on the sample and properties of hypothesis class.…”

“…The structured query language (SQL) is created to manipulate, store and query relational databases [9], [10]. SQL is carried out by a leading database management system data base management system (DBMS) and used in a wide variety of sectors [11].…”

Towards online training for RL-Based Query Optimizer

“…One of the primary challenges in database research trends is query optimization using reinforcement learning models [4][5][6][7][8][9][10][11]15]. To enhance query optimization, researchers have explored employing Reinforcement Learning approaches.…”

“…Recently, a Deep Reinforcement Learning Based Query Optimizer (RL QOptimizer) has been presented [4]. The proposed method utilizes reinforcement learning and real-time feedback from the database to determine the best join order in query plans.…”

“…As a result, most optimizers use heuristics to reduce the search space, such as examining the structure of the query tree [3]. Recently, reinforcement learning and its reward system [4][5][6][7][8][9][10][11] were utilized to learn prediction models for the best join orders between tables over a snapshot of the database. These approaches used reinforcement learning, where the agent receives continuous feedback on actual response time or estimated cardinality, rather than relying on pre-calculated statistics, to maximize the reward, which results in improved query performances.…”