Automating Oracle Database Performance Tuning Through AIBased Workload Analysis

Abstract

Oracle databases effectiveness in enterprise applications is significant. However, the tuning process is time-consuming, and it requires profound knowledge. Static tuning strategies employed in previous times are inadequate when it comes to working with fluctuating loads and changing application topologies. This paper proposes a well-articulated plan for Oracle database performance tuning using AI workloads auto evaluation. This proactive identification is achieved through performance prediction using Machine Learning (ML) algorithms, the characterization of workload, and performance history data from the baseline analysis. This paper focuses on system formulation, data accumulation, feature extraction, model development, and decision-making processes. Overall, the effectiveness of the proposed framework was thoroughly evaluated using real clustering enterprise workload, resulting in significant achievements in query execution time, resource consumption, and system instructiveness. This paper also outlines how the system can be customized on how it works depending on the workload placed on it and its performance compared to that of the rule-based tuning method. The experiment's findings prove the proposed idea of the AI-based tuning approach providing performance improvement while simultaneously eliminating administrative burden. This is where the proposed solution shows that it is possible to overcome the increased database complexity and its requirements for high availability, scalability, and performance through intelligent automation.